Repository path: workshop/translations/maqamat-hulwan/collation.md · rendered 2026-09-09

Copy-text and collation — «المقامة الحلوانية»

The copy-text

ar.wikisource.org, «مقامات الحريري/المقامة الحلوانية», fetched 2026-08-26 via action=raw (HTTP 200, 10,302 bytes). Public domain: al-Ḥarīrī died 1122. The transcription is fully vocalised and punctuated at every saj' colon boundary, exactly as the same site's «المقامة الصنعانية» is, which is what makes the 140-colon segmentation in source-arwikisource-maqama2.txt a reading of the page rather than a judgement of the lead's.

Stored as source-arwikisource-maqama2.txt, one record per line, ID<TAB>FORM<TAB>text. 140 prose cola and 17 verse lines, 709 Arabic tokens (532 of them prose). Three trailing lines of MediaWiki category markup were dropped and that is the only edit.

The collation, and why there is not one

This copy-text is single-witness, and the same four routes that failed for «المقامة الصنعانية» two days ago fail here for the same reasons — wiki/base/anchors/A-hariri-hands §1 and ../maqamat-sanaa/collation.md record them: the ar.wikisource API returns HTTP 429 to this container, both archive.org Arabic printings have unusable OCR of lithographed naskh, and al-maktaba.org returns HTTP 403. They were not re-attempted; a route that failed on Assembly I two days ago is not a different route on Assembly II.

Both nineteenth-century printings in this session's hands print the Arabic alongside their English, and neither was opened — for the reason ../maqamat-sanaa/collation.md gives: a witness bought at the cost of the study's independence is not a witness. Here the cost would have been larger still, because this session's whole question is what those two printings do.

So the limit is declared as it was for «الصنعانية»: the Arabic of this maqāma rests on one transmission. Nothing in RS-20260826-balanced-period turns on a disputed Arabic reading; what it measures is the length and pairing of English clauses, and the Arabic enters only as the per-colon token counts of §5's source-tracking control, where a single variant word would move a count by one.

What the copy-text's shape already tells the study

The Arabic cola of this maqāma are markedly uneven where the story turns. The verse-introducing formulae — «وأنْشَد» and he recited, «فابتَدَر» then there sprang forward — are one and two words standing next to neighbours of six and seven. Any English hand that follows the copy-text's colon boundaries inherits that unevenness, and any measure of "balance" computed over consecutive cola will be dragged by it. That is why RS-20260826-balanced-period computes the same statistic on the Arabic and reports the English against it, rather than against zero.