Bench&Brace

Policy

Sourcing and Accuracy Policy

Where the information on this site comes from, how confidence is rated, what the review gate enforces, and how corrections are handled.

Accuracy matters more in this subject than in most. The people reading a dating guide are about to spend money on the strength of it, and they know the material well enough to catch a mistake immediately.

This page sets out what that means in practice.

Where information comes from

Sources fall into four tiers, and every record on the site names the ones behind it. The full list, with reliability ratings and copyright status, is on the sources page.

Primary documents — manufacturer catalogues, patent filings, trade directories. The strongest evidence available. A catalogue proves a model was offered in a given year at a given price. A patent proves a mechanism existed by a given date.

Standard reference works — the published books and reference sites that the hobby has cross-checked for decades. These are usually right and are the practical backbone of any tool reference, but they are secondary sources and they do occasionally propagate an error between them.

Community identification — forums and enthusiast archives. Genuinely valuable for spotting variants nobody catalogued, and the right place to find out that something unusual exists. Never sufficient on its own for a date.

Market data — filtered sold-listing data for value ranges. Noisy, biased, and the only public sample of realised prices there is.

What the build refuses to publish

The review gate is enforced by the build, not by good intentions. A record cannot carry published status unless it has:

  • a substantive summary, not a one-line caption
  • at least two concrete identification points
  • a production start year
  • at least one cited source
  • no outstanding verification items

A record failing any of those is demoted to draft automatically. Draft pages still build — hiding them would hide the gaps — but they carry a banner saying they are unverified, they list what still needs checking, and they are excluded from search engines with a noindex directive.

At the last build, 58 of 86 model records were published and 28 were drafts. That ratio is deliberate. Publishing a thin page with a photo and a guess is how a tool reference loses its audience.

How confidence is rated

Facts on this site carry one of three confidence levels, shown in the interface rather than buried:

  • High — attested by a primary document, or by multiple independent standard references in agreement.
  • Medium — from a standard reference, uncorroborated, and consistent with everything else known.
  • Low — plausible, in circulation, not yet verified against a source anyone can check.

Type-study features carry confidence individually rather than as a whole, because the sequence is much better attested in some places than others. The 1885 lateral lever and the 1902 toe marking are near-certain; the distinctions between the earliest types are much softer, and the site says so on the page rather than in a footnote.

Why the dates are ranges

Stanley never published a type study — it was reconstructed by collectors from catalogues, patent dates and thousands of observed planes. The type numbers are a collector's convention.

More importantly, the underlying reality is fuzzy. Stanley emptied parts bins before opening new ones, so a plane assembled in 1911 can legitimately carry a 1908 casting and a 1912 iron. Any source giving you a single year for a Bailey plane, from the plane alone, is overstating what the evidence supports.

How value ranges are built

Every value block states its currency, an as-of quarter, an approximate sample size and its method. Bands come from manually filtered sold listings with reproductions, obvious marriages and mis-titled listings removed.

Where the sample is thin, confidence is marked low and the band is wide. A model with fifteen recorded sales a year does not have a tight market price, and presenting one would be inventing precision.

These are researched estimates for a private sale, not appraisals. For insurance, probate or a significant collection, use a specialist auction house.

Images and rights

Photographs of casting marks are essential to this subject and are also the easiest way to infringe someone's copyright.

The rules here: US patent documents carry no copyright and can be reproduced freely. Catalogues published in the US before 1930 are public domain, though an individual scan may carry terms from the institution that made it — those get checked per scan. Community-contributed scans are linked, never rehosted. Auction listing photographs are never reproduced. Photographs commissioned or taken for this site are marked as such.

Any image slot on this site that is still empty is shown as an empty slot with a description of the photograph needed, rather than filled with something whose provenance has not been established.

Corrections

Corrections are welcome and get made. The most useful form is: the page, the specific claim, and the source that contradicts it.

Where a correction changes a published claim, the record's source list and confidence rating are updated with it, so the reasoning stays visible rather than just the conclusion changing.

Sources