Repository path: workshop/canon/niewola-tatarska/manifest.md · rendered 2026-09-09

Source manifest — Sienkiewicz, «Niewola tatarska», section I (opening)

Work. Henryk Sienkiewicz, Niewola tatarska. Urywki z kroniki szlacheckiéj Aleksego Zdanoborskiego (1880) — "Tartar Captivity. Fragments from the chronicle of the nobleman Aleksy Zdanoborski." A first-person pastiche of a seventeenth-century Polish nobleman's memoir, macaronic with Latin, in archaic orthography.

Public domain. Sienkiewicz died 1916.

Language. Polish, with embedded Latin. The project's first Polish source, and its first source that is macaronic by design.

Witness

Single witness, and that is a limitation declared here rather than discovered later.

No collation was possible and none is claimed. Wolne Lektury's text preserves nineteenth-century orthography (téż, daléj, tém, wonczas, końskiemi, szlacheckiéj) rather than modernising it, which is evidence that it follows a period edition and is not a silent modernisation; that is an inference from the text's own surface, not a statement about which edition. A second witness would be worth having and the project has not found a free one.

The unit translated

Section I, paragraphs 1–2 entire, plus the opening of paragraph 3 cut at a sentence boundary.

• Extraction: characters 119–19,273 of the decoded file are section I; that region splits on blank lines into the section number and nine body paragraphs. Paragraphs 1 (262 words) and 2 (54 words) are taken entire.
• Paragraph 3 is 1,632 words with no internal newline — the source's own paragraphing, not an export artefact (asserted: count('\n') == 0 inside it). Taking it whole would have made the unit 1,949 words; taking none of it would have left the unit at 316. It is therefore cut at a sentence boundary, after the twenty-sixth sentence, ending:

…Jakoż tam jest najlepsza rycerska szkoła i zaprawa.

The cut is declared because it is arbitrary in a way a paragraph boundary would not be. It falls at the end of a completed movement — the narrator's summary of the borderland as a school of arms — and nothing syntactic is broken.

source.txt: 985 words, 6,179 characters, three paragraphs. sha256 7aa9b4912f27e63887b08d1e0391a7676e885b97a33fb54827d34e77b5e3c2d1.

Units, for the standing contamination selection gate: unit A = paragraphs 1–2 (316 words); unit B = the cut portion of paragraph 3 (669 words).

Comparator (for the contamination gate and for ARM-revision step 2b)

Jeremiah Curtin, "Tartar Captivity," in Hania (Little, Brown, 1898) — Project Gutenberg #36583, https://www.gutenberg.org/ebooks/36583.txt.utf-8, fetched 2026-07-30, HTTP 200, 887,414 bytes, sha256 22fc49837d9a536b0e5b979ca58109b18719f72271c417bffdf06357f7023720.

Located by heading offset only; no prose displayed to the translator (method note abm). The heading TARTAR CAPTIVITY. is at character offset 270,408; the next story's heading (LET US FOLLOW HIM) at 348,015; the story region is 77,607 characters / 14,189 words. The comparator window used by tools/dependence_check.py is the first 3,000 words of that region — deliberately over-long, roughly triple the passage, because over-extraction can only make a contamination check more likely to fire and never less. sha256 81bcc98dc6042c81986a148643805ff65fb084b45c2b44dc1b597b7cd6d44fc3 on the window.

Neither the volume nor the window is stored in this repository. Every figure that uses it is reproducible from the URL, the digest and the two offsets above; tools/dependence_check.py reads it and reports counts, and the translator never opened it.