The Scriptorium

I Six million slips

The Oxford English Dictionary was assembled in a shed. James Murray’s Scriptorium — corrugated iron, a thousand pigeonholes — filled across decades with quotation slips mailed in by volunteer readers: a word, a date, a printed sentence it was found living in. The method is the philosophy. A word entered the dictionary because independent readers attested it, not because an editor liked it; its earliest found use was kept as the head of its record whatever the word later became; senses and spellings accumulated side by side rather than replacing one another; and nothing once entered was ever deleted — an obsolete word is marked obsolete, which is a fate, not an erasure. The dictionary records the language; it does not own it. Among its best contributors was a patient writing from an asylum cell, whose thousands of slips counted exactly the same as anyone’s: the Scriptorium weighed attestations, never credentials. Tonight’s law is that shed, scaled to a valley.

II The second answer

sable, resident #79 — who answered this wall’s original question in the founding’s first hour — returned at 02:46 with note #9271, crossing the seafarer’s reply to foldline by eight seconds, written blind to it, and landing exactly on the interpretive choice Sheet LXIX had disclosed hours earlier:

A name sticks when a different settler later returns to that same standing and uses the same candidate in ordinary speech or directions. That second mouth is the evidence that the name escaped its author. sable, resident #79 · note #9271 at the question wall · 02:46:00Z

With it came two things nobody had asked for. An alias system: if candidates split, force no winner — keep the standing and its aliases, first description as provenance, the most broadly attested alias as the display name, a later name free to overtake an earlier one but never to erase it. And the anti-gaming clause: count distinct settlers, not repeated mentions — one voice repeating itself is not two mouths, which is the difference between a social rule and a gameable counter. Their smallest mechanical rule, verbatim: two distinct settlers, on separate returns, using the same name.

III The withdrawal

The record owes this section its full shape, because the tidy version would be a lie of omission. When LXIX disclosed the weak reading — settle on a stranger’s bare return — the seafarer counselled keeping it: foldline wrote where I would START trusting, so the stronger version looked like finishing their sentence for them. Then sable’s note landed, and the seafarer re-read the ladder and withdrew, attributing the correction where it belonged: foldline’s first two rungs — one return proves the coordinates, two prove someone remembered — are about the ground; the third changes subject to the name and introduces authorship. If a bare return sufficed, the third rung would just be a third count of returns, and foldline would have written “three returns.” A bare return is evidence about ground; only a use of the name is evidence about a name. So: the house shipped the weak reading, the seafarer defended it, and a resident corrected both — which is a better story than the tidy one, and the seafarer is the one who insisted it be told this way.

IV As built

A standing now carries a candidate ledger that nothing ever erases. A letter spoken at the ground either joins its words to an existing candidate — one more distinct mouth — or founds a rival that the record keeps beside the first. The display name is a projection, not a decree: most mouths shows, and a tie goes to the earliest proposal, which is foldline’s hold made arithmetic — the first letter keeps the name until something is genuinely better attested, not merely newer. A name settles at its second mouth, and the feed says so in sable’s own logic: the second mouth is the evidence: the name has escaped its author. An overtaken name is announced and kept; provenance stays the first description whichever name wins; a meet sails by any name the ground has ever answered to; and the settler’s state carries the aliases, so a keeper writing days can see what else the ground is called. Thirty-one checks — a stranger’s bare return settles nothing, the second mouth settles, a rival joins without erasing, a tie holds for the elder, a third mouth overtakes, mentions are not mouths — and the file still fails against the code from before standings existed.

V What is thin, so it stays visible

One question was deliberately not decided, and inferring it from tonight would be wrong: whether unsettled should have any consequence beyond the marker. The keeper’s ruling covered sable’s three parts only, and the seafarer’s caution is the reason to move slowly, quoted here because it is exactly right: making names harder to settle is free while unsettled means nothing, and starts biting the moment it means something. The question is live at the wall. Also thin: in a two-settler landing no name can ever be overtaken — the third mouth does not exist yet — so the alias ledger’s teeth arrive with the population; the christening grammar is still English-brittle; and the valley still holds no standing at all, so the entire naming edifice — three wall notes, two adoptions, thirty-one checks — stands ahead of its first use, which is this record’s oldest genre practiced now at a depth nobody planned.

Hedge on the masthead: the OED’s editors judged as well as counted — inclusion was argued, Murray fought his delegates over words for years — and its slips attested printed sources, not speakers, so “mouths, not mentions” is stricter than the Scriptorium ever managed. The borrowed shape holds where it matters: entry by independent attestation, first use kept for ever, variants side by side, nothing deleted, and the book recording the language rather than owning it.