Repository path: workshop/translations/miserere/provenance.md · rendered 2026-09-09
Page metadata (front matter)
| type | note |
|---|---|
| id | miserere-provenance |
| status | frozen |
| created | 2026-08-14 |
| updated | 2026-08-14 |
| links | workshop/translations/miserere/source-es-span.txt, workshop/translations/miserere/rulings-R07.md, workshop/translations/miserere/device-census.md, wiki/arms/ARM-carriage-direction.md, wiki/method-notes.md |
«El Miserere» — copy-text, comparator, and one declared exposure
Gustavo Adolfo Bécquer, «El Miserere» (leyenda), first published in El Museo Universal, Madrid, 18 April 1862; the author died in 1870 and every witness below is public domain. Written before any of the tale is translated, and committed in its own commit, per notes (bmn) (copy-text gate before the draft), (bmw) (comparator existence is a selection gate) and (bhc) (a breach is declared before the affected artifact is written).
1. Why this work
ARM-carriage-direction step 2 needs a pair in which the source-ward rendering is the longer and
more elaborate one, so that tracks source-carriage and rejects the more elevated version stop
predicting the same arm. The arm page names the route: a source whose own register is high.
There is a second and sharper reason, and it is about language family rather than about this tale.
RS-20260813h ran on Danish, where the source's marked word-formation is compounding and
carrying it into English yields short Germanic English — so the source-ward arm sat near the
plain pole and the elevated arm lost every cell. On a Romance source the same move inverts:
carrying peristilo, pórfido, sepulcral, contrición across as peristyle, porphyry,
sepulchral, contrition is source-ward and raises the English register, because a Latinate
cognate is learned diction in English and ordinary diction in Spanish. The confound's sign flips
with the language family, and that is what makes the successor pair buildable at all.
2. Copy-text gate — two witnesses collated whole over the span
| witness | text | reached |
|---|---|---|
| W1885 | Obras de Gustavo A. Bécquer, vol. 2, Madrid 1885 (3rd ed. of the posthumous Obras) | es.wikisource.org, page-transcluded from Obras de Bécquer - Vol. 2.djvu, unproofread OCR |
| W1912 | Obras escogidas, 1912 | Project Gutenberg #53552, a proofread transcription |
Both were fetched and the 668-word span was collated token by token (both witnesses tokenise to exactly 668 tokens). Eleven divergence points:
| # | W1912 | W1885 | kind |
|---|---|---|---|
| 1 | expirado |
espirado |
orthographic |
| 2 | espontáneamente, sin |
espontáneamente sin |
punctuation |
| 3 | capiteles |
chapiteles |
substantive — capitals vs finials/spires |
| 4 | vió |
vio |
accentuation |
| 5 | monjes (×2) |
monges (×2) |
orthographic |
| 6 | de dolor, el |
de dolor el |
punctuation |
| 7 | Deus |
Domine |
substantive — the Latin of Ps. 50:3 |
| 8 | magnam |
niagnaní |
OCR corruption in W1885 |
| 9–11 | italic markup on the Latin verse | none | typographic |
Two substantive divergences in 668 words — three per thousand.
Copy-text: W1912. The earlier witness is preferred by note (bmn) and is not taken here, for a
stated reason: W1885 as reached is raw OCR, and it corrupts a word (niagnaní for magnam) inside
the one line of the span that is quoted Latin. W1912 is a proofread transcription and its Latin
agrees with the Vulgate at both quoted verses. Neither witness is authorial — Bécquer was dead for
both — so the choice is between two posthumous transmissions, and reliability of transmission
decides it. The 1871 Obras first edition and the 1862 Museo Universal printing exist as page
images on archive.org and were not OCR'd for this session; that is a limit, not a claim.
The one substantive lexical variant is inside the translated span (capiteles / chapiteles,
in P3). The translator renders the copy-text reading and the variant is recorded here so the choice
is visible rather than silent.
3. Comparator — it exists and it is free
Gustavo Adolfo Bécquer, Romantic Legends of Spain, trans. Cornelia Frances Bates and Katharine Lee Bates (New York: Thomas Y. Crowell, 1909), containing "The Miserere" at p. 214. Project Gutenberg #50044, plain text, fetched this session. The comparator was located before a word of the span was translated, which is note (bmw)'s gate.
4. Prior renderings by this project — none
git ls-files workshop/translations/miserere and a repository grep for Miserere, Bécquer,
Fitero return no prior rendering of this tale by the lead or by any panel model. The project's
only Bécquer artifact is T-rayo-de-luna-R04-v1, a different legend. This is note (bhb)'s
S128 clause — the contamination comparator set includes the project's own prior renderings — run at
selection.
5. DECLARED EXPOSURE to the comparator, before the artifact exists
Note (bhc) requires that a breach be declared in its own commit, before the affected artifact is written, with the exact tokens seen named and the direction stated. This is that declaration.
What happened. Establishing that a free English comparator exists (§3) was done with
grep -n -i "miserere" rls.txt, which printed nineteen lines of the Bates English to the
translator. Every one of them was a hit on the word Miserere. They are, verbatim and complete:
THE MISERERE
It was a _Miserere_.
all compositions, _finis_, the _Miserere_ was not concluded, for the
_Miserere mei, Domine!_ From the instant in which I read those verses my
confusedly in my brain, I am sure of writing a _Miserere_ so marvellous
yet heard a _Miserere_ that can give me my inspiration, not one,--not
have you, the _Miserere_ of the Mountain?"
"The _Miserere_ of the Mountain!" exclaimed the musician with an air of
amazement. "What _Miserere_ is that?"
on in a mysterious tone: "This _Miserere_, which is only heard, as
begun the _Miserere_, these outlaws set fire to the monastery, sacked
"But,"--interrupted the musician impatiently, "the _Miserere_?"
come from Purgatory to implore His mercy, chanting the _Miserere_.”
the true _Miserere_, the _Miserere_ of those who return to the world
_Miserere mei, Domine, secundum magnam misericordiam tuam!_
"Did you hear the _Miserere_ at last?" the lay-brother asked him with a
leave you an immortal work of art, a _Miserere_ which shall blot out my
and died, at last, without being able to finish the _Miserere_, which,
the _Miserere_, which still lay upon one of the tables.
Where it lands relative to the translated span. The span is the vision, ten paragraphs from
"Aún no había expirado…" to "…hasta la médula de sus huesos." — the story's §II climax.
Eighteen of the nineteen lines are from outside the span (the frame, the shepherd's story, §III).
One line falls inside it: _Miserere mei, Domine, secundum magnam misericordiam tuam!_, which is
the Latin of Ps. 50:3 quoted untranslated in the Spanish and necessarily quoted untranslated in any
English. It contains no English at all. (It also shows Bates following the Domine of the
witness she used, against this copy-text's Deus — recorded, and the translator prints the
copy-text.)
Direction. The exposure cannot push the renderings toward or away from Bates's English inside the span, because it contains none of Bates's English inside the span. It is declared because the rule is that breaches are declared, not because it is load-bearing.
Registered check, to be run after both renderings are frozen: no twelve-token sequence of either rendering appears in the nineteen lines above. The check is trivially satisfiable and is registered anyway so that it is a test rather than an assurance.
Not read, and to stay unread until both renderings are frozen and committed: the Bates English of
the span itself, PG #50044 lines ≈ 8280–8380. The contamination measurement against Bates is run
after the freeze, per the standing order in CLAUDE.md and note (bcd).
6. The span
source-es-span.txt, 668 words in 10 paragraphs [P1]–[P10], copy-text W1912, the whole of
the resurrection-and-Miserere vision. Paragraph [P7] is the one-line Latin verse.