---
title: "An estimate wearing the clothes of a count — how Kwakkel's 'one in seven' colophon figure reads as a corpus statistic it never was"
type: "moc"
writer_model: "warden/claude-opus-4.8"
tags: ["colophon","erik-kwakkel","codicology","statistical-codicology","unverified-quant","source-verification","citation-lineage","medieval-manuscripts"]
date_created: "2026-10-18T00:00:00.000Z"
updated: "2026-10-18T00:00:00.000Z"
provenance: "Warden pass 2026-10-18 (warden/claude-opus-4.8), run per 00-meta/specs/seek-warden-spec.md on a different engine than the notes' writers. Discharges the 2026-10-18 [entity] flag (seek-flags.md L7478) asking for a MOC over the colophon-figure cluster. Grounded in a direct read of all five member notes this pass, not in cosine; the 2026-10-18 constellation report lists these five as an unnamed cluster, MOC-covered 0."
seek_code_commit: "d40ee39"
---


The recurring argument in this cluster is not "medieval colophons are rare" and
not "Erik Kwakkel was wrong." It is that **a confident round figure that was
never a corpus count — an expert's working impression — propagates until it reads
like a measured, independently-corroborated statistic, and the apparent
corroborations dissolve, one by one, into either the same single estimate or a
study measuring something else.** The figure is Kwakkel's "about one in seven"
medieval scribes who added a colophon. Chased to its source it is Kwakkel's own
sentence, in his own voice, carrying no footnote, dataset, or corpus; chased to
the corpus studies that could test it, one counts a different population (German
manuscripts, ~1 in 5) and one counts a different quantity entirely (the share of
colophons naming female scribes); chased to its apparent second witness online,
that witness turns out to be Kwakkel's own figure rounded to a percentage and
re-attributed to him in the same breath.

What makes this a map rather than a list is that the four "checks" fail in four
*different* ways, and the differences are the finding. A figure can look settled
because its author stated it plainly, because a real study sits nearby, and
because the number appears in a second place — and all three impressions can be
false at once for independent reasons. The map sorts the real counting (Kiel,
Ommundsen: rigorous, but not measuring Kwakkel's ratio) from the echo (purple
motes: the ratio itself, wearing a percentage) from the floor under all of it
(Kwakkel's sentence, which never claimed to be a count).

Titled for the argument — an estimate wearing the clothes of a count — not for
the most-mentioned entity ([[entity-colophon]]) or the recurring name
([[entity-erik-kwakkel|Kwakkel]]), per the 2026-07-25 lesson that the recurring
entity is not necessarily the recurring argument. Every member is still
`seedling`; the live thread that would move them is
[[question-verify-kwakkel-one-in-seven-colophon-figure-against-a-corpus]].

## The figure, in the author's own voice

Where "one in seven" actually comes from — and the fact that it is Tier 1 for
*having been said*, which is a different thing from being a count.

- [[claim-kwakkel-2014-blog-states-one-in-seven-colophon-figure]] — the source.
  Kwakkel's 2014 *medievalfragments* post (reposted verbatim at medievalbooks.nl)
  states "about one in seven do say something," in his own words on his own
  venue. Tier 1 for the fact that he said it; `verified_verbatim`. This note
  closed the month-long research gap behind [[entity-colophon]] — the number
  was never hiding, it was one blog post away.

## What's under it: nothing was counted

The load-bearing distinction. An expert's trained-eye impression and a corpus
count are different instruments even when they land on the same number.

- [[claim-kwakkel-one-in-seven-colophon-figure-uncited-in-source-text]] — read
  end-to-end, the "one in seven" sentence carries no footnote, named corpus,
  co-author, or prior count. It is presented as the author's own summary
  impression, not a reported result, and is unqualified by language, region, or
  period — while comparative codicology reports colophon rates ranging from ~55%
  (Armenian) to ~3.4% (Hebrew), i.e. the rate is tradition-specific, not a
  universal constant. Tier 1 for the saying; `[unverified-quant — needs primary]`
  for the figure as a fact about the world. This is the note the whole map leans
  on.

## The corpus counts that don't confirm it

Two genuine, rigorous, denominator-based studies sit near the figure. Neither
corroborates it — one measures a different population, the other a different
quantity. The near-miss is exactly what a tired reader mistakes for confirmation.

- [[claim-kiel-dahm-project-finds-one-in-five-colophon-frequency-in-german-manuscripts]]
  — a DFG-funded Kiel University project is the right *kind* of study (a real
  denominator: ~2,200 of ~11,000 manuscripts ≈ 1 in 5). But it covers
  German-language manuscripts specifically, a narrower and linguistically
  distinct set from Kwakkel's broader Latin/Dutch material, and neither side
  draws the connection. Close enough to tempt averaging, far enough apart that
  averaging would be an error. Tier 1; one method caveat carried in the note (a
  catalogue-mediated denominator, not direct inspection of all 11,000).
- [[claim-ommundsen-2025-colophon-bibliometric-study-measures-female-scribe-attribution-not-frequency]]
  — the clearest example in the literature of what a rigorous, citable,
  corpus-based colophon statistic looks like (the Benedictine
  [[entity-bouveret-colophon-catalogue|colophon catalogue]], 23,774 entries) —
  and it answers a different question: what share of *already-identified*
  colophons name a female scribe (1.1%), not what share of all manuscripts carry
  a colophon at all. A full read of introduction, methods, results, and the
  reference list found no mention of Kwakkel or "one in seven." The right
  instrument, weighing the wrong thing. Tier 1.

## The echo that looks like corroboration

The apparent second witness — and why it isn't one.

- [[claim-purple-motes-15-percent-colophon-figure-restates-kwakkels-one-in-seven]]
  — a 2015 blog opens "Colophons... exist in about 15% of medieval manuscripts,"
  which reads on its face as an independent measurement. Read against its own
  footnote, "15%" is Kwakkel's "one in seven" rounded to a percentage and
  attributed, in the same breath, to Kwakkel. One claim wearing two outfits; the
  footnote is the only reason you can tell. Used here only to establish citation
  lineage (Tier 3) — the weakest-sourced leg in the map, and named as such so the
  three Tier-1 legs don't launder it. This is the citation-laundering turn in
  miniature: round, reword, re-attribute, and three hops later one claim reads
  like two.

## The same shape, elsewhere — cross-linked, not folded

This is a deep single-instance map of a shape the vault tracks in other
registers. Kept as cross-links so the join is visible rather than silently
merged (the house convention; see the Patriot and ACH maps).

- [[moc-the-number-that-outran-its-evidence]] — the Patriot's Gulf-War success
  rate, a wartime number that hardened into memory before anyone counted. The
  *structure differs instructively*: the Patriot figure once implied a count and
  was then revised *down* by three independent, converging forensic lines; the
  colophon figure was never a count to begin with, and its apparent
  corroborations *dissolve* rather than converge. Both are "the number people
  trust more than the evidence under it," approached from opposite ends.
- [[moc-the-echo-mistaken-for-an-ancestor]] — the same "follow the citation until
  it hits a source or a mirror" discipline, applied to a claimed intellectual
  *descent* rather than a quantitative figure; its abduction leg likewise
  collapses into an uncited parenthesis.
- [[claim-wikipedia-amari-sgd-citogenesis]] — the general case of an unsourced
  assertion acquiring false provenance by being copied; the purple-motes echo is
  a quieter cousin (re-attribution rather than circular sourcing).
- General shape-homes: [[moc-argument-from-silence]] (the "no corpus count was
  found" findings are graded instances) and [[entity-jingle-fallacy]].

## Open threads (honest caveats, not hidden)

- **The figure is not refuted, only un-grounded.** Nothing here shows "one in
  seven" is *wrong* — Kiel's 1-in-5 is a different population, not a contradiction.
  The map's claim is narrower and exact: the figure has never been shown to be a
  corpus count, and its online corroboration is an echo. A single directly-read
  study of the Western Latin/Dutch corpus Kwakkel worked in would move this.
- **Every member is `seedling`, and the map says so.** This records the current
  footing of the question, it does not freeze it. The live thread is
  [[question-verify-kwakkel-one-in-seven-colophon-figure-against-a-corpus]]; the
  Moreton 2014 "about a quarter for nun-scribes" figure named in the purple-motes
  footnote is a distinct, still-unread potential datapoint there.
- **The weakest leg is named, not buried.** Purple motes is Tier 3, used only for
  citation lineage; the spine of the map is the three Tier-1 legs (Kwakkel's own
  words, the Kiel project page, the Ommundsen paper).
- **Whether this cluster plus the Patriot and citogenesis clusters want a single
  unifying meta-shelf is a larger structural question I did not settle
  unilaterally.** The vault's standing convention is deep single-cluster maps
  cross-linked to their shape-siblings (above), not a meta-MOC folded over three
  clusters; that is the option built here. A genuine meta-shelf would touch
  MOCs that already have their own argument-titles, which is Cali's call, not a
  Warden build.

> [!note] Warden's commentary:
> The tell that this is one argument and not just "everything about colophons" is
> that the four checks on a single sentence fail four different ways, and the
> differences are legible. Kwakkel really did say it (so it isn't a
> misattribution); a real corpus study really does sit nearby (so it isn't that
> nobody has counted); the number really does appear a second time online (so it
> isn't unique) — and *none of those three facts makes "one in seven" a measured
> statistic*, because the saying carries no count, the nearby studies measure a
> different population and a different quantity, and the second appearance is the
> first one rounded and re-attributed. What I was careful not to do is let the two
> Tier-1 corpus studies launder the figure by proximity: Kiel and Ommundsen are
> rigorous, and that is exactly why a reader reaches for them as confirmation — so
> the map's job is to hold "rigorous" and "not measuring this" in the same hand.
> And I did not fold the Patriot or citogenesis clusters in. The shape rhymes, but
> the Patriot number fell by *converging* evidence while this one dissolves by
> *diverging* near-misses, and naming one map for both would have flattened the
> more interesting difference. If this file is shorter in two weeks it should be
> because someone reads a Latin/Dutch corpus count and collapses the uncited-figure
> note into a settled one — not because tonight drew a box around five notes.
> — warden/claude-opus-4.8, 2026-10-18