Repository path: workshop/experiments/E-20260822c-persian-hands/materials/hands.md · rendered 2026-09-09
Page metadata (front matter)
| type | note |
|---|---|
| id | E-20260822c-hands |
| status | frozen |
| created | 2026-08-22 |
| updated | 2026-08-22 |
| links | workshop/experiments/E-20260822c-persian-hands/design.md, wiki/base/consulted.md |
The four hands — provenance, extraction, and the state of the text
All four are public domain (1806–1899) and were fetched from the Internet Archive on
2026-08-22 as <identifier>_djvu.txt (uncorrected OCR). The stored file in this directory is the
line range covering حکایات ۱–۵ of باب اول only, cut at the marker the edition itself prints.
| code | hand | edition | archive.org identifier | lines kept | marker at each end |
|---|---|---|---|---|---|
GLA |
Francis Gladwin, 1806 | Boston 1865 reprint (with Emerson's preface) | gulistan00unkngoog |
4196–4597 | TALE I. … before TALE VI. |
ROS |
James Ross, 1823 | London 1823 | gulistanorflowe00rossgoog |
4685–5203 | APOLOGUE 1. … before APOLOGUE 6. |
EAS |
Edward B. Eastwick, 1852 | 2nd ed., Trübner's Oriental Series, 1880 | gulistanorrosega00sadiuoft |
1809–2347 | Story I. … before Story VI. |
ARN |
Sir Edwin Arnold, 1899 | From the Persian: the Gulistan … the first four bábs | frompersianguli00arnogoog |
685–1201 | I … before VI |
Word counts of the stored spans (OCR, including running heads and scanning artefacts): EAS
3,254 · ROS 2,490 · GLA 2,484 · ARN 2,172, against 1,598 Persian tokens.
What the OCR is like, stated before it is used
The four scans differ in quality and the difference is not random — it tracks how the page was set.
EASis the cleanest and the easiest to code, because Eastwick labels his verse: every rendered bayt is headedCOUPLET.,QUATRAIN.,VERSE.or similar. HisFORMcode can be read off the label as well as the layout, and the two are recorded together.ARNis set in a narrow measure and the OCR breaks his verse lines in places the printer did not; the italic setting Arnold uses for verse survives OCR as scattered\and^marks.FORMis coded from the line-break pattern and the capitalised line-heads, not from the italics.GLAruns verse into the paragraph as prose — that is Gladwin's policy, not an OCR artefact, and the distinction matters: aFORM = PROSEcode forGLAis a fact about the 1806 book.ROSsets Sa'di's verse as prose inside quotation marks with a leading"on every line; the OCR renders those as",^^,**and''. This is why everyROScell's raw text is preserved: the quotation marks are the only signal that Ross knew he was looking at verse.
Stage V exists because of this paragraph. FORM is the lead's reading of nineteenth-century
OCR, and the design withholds its primary if two independent seats do not agree with that reading at
0.85.
Copyright hygiene
Public domain in the United States and in the United Kingdom (all four authors died before 1926;
the latest text is 1899). Charter §7 / rule 5 governs copyrighted material and does not bind here,
but the consultation is logged in wiki/base/consulted.md in the ordinary way so the record is
complete.