Repository path: workshop/translations/kalila-saih/collation.md · rendered 2026-09-09
Page metadata (front matter)
| type | note |
|---|---|
| id | kalila-saih-collation |
| status | frozen |
| created | 2026-08-16 |
| updated | 2026-08-16 |
| links | workshop/translations/kalila-saih/R30-v1/translation.md, workshop/translations/kalila-saih/source-wikisource-saih.txt, workshop/translations/kalila-saih/witness2-amiriyya-1937.txt |
Copy-text and collation — «باب السائح والصائغ»
1. The witnesses
| witness | provenance | state | |
|---|---|---|---|
| copy-text | Arabic Wikisource, كليلة ودمنة/باب السائح والصائغ |
no edition statement; unvocalised; modern orthography | clean text, 998 words of body, 60 sentences |
| witness 2 | Arabic Wikisource, كليلة ودمنة (الأميرية، 1937)/باب السائح والصائغ |
Cairo: al-Maṭbaʿa al-Amīriyya (Būlāq) 1937, pp. 301–307, page images present | vocalised; optical recognition heavily damaged |
A third candidate was searched for and is not usable. Louis Cheikho's La version arabe de
Kalîlah et Dimnah d'après le plus ancien manuscrit arabe daté (Beirut 1905, archive.org
laversionarabed00cheigoog) is the oldest dated manuscript in print and would be the better second
witness; its scan carries no recoverable Arabic text at all — a search of its 246 KB of
recognised text returns zero occurrences of السائح, الصائغ or ركية. Recorded so the search is
not repeated.
2. The collation
Whole-chapter, word by word, after stripping vowel marks and normalising alif/yāʾ/tāʾ marbūṭa. 1,001 normalised words against 1,036; 68 non-identical spans; sequence ratio 0.9229.
Sixty-one of the sixty-eight are optical-recognition noise of a single, systematic kind: the
1937 transcription loses the dots that separate ج/ح/خ, غ/ع, ب/ت/ي and ف/ق. الصائغ → الصائع
(eight times), الببر → البير (four), الجن → الحن, جحرها → خرها, يختصوا → يحتصوا,
لدغت → لدعت, والفاجر → والفاحر. Five further spans are page furniture pulled into the text
(موسو, مرموش, البساتين, مه مهمه مر, the running head).
Two substantive readings adopted from witness 2, and both are corrections to the copy-text:
| sentence | copy-text | witness 2 | adopted | why |
|---|---|---|---|---|
| 8 | فإن الطيب الرفيق العاقل |
فإن الطبيب |
الطبيب | the clause is about examining a patient, feeling his veins and diagnosing his illness; the good man cannot do it and the physician is the standard reading |
| 45 | فدعى الملك أهل العلم |
فدعا |
فدعا | the copy-text's فدعى is an orthographic slip for the III-weak perfect |
Two readings inspected and NOT adopted, recorded because they were live: witness 2's
ولا يصطنعوا أحداً (plural, agreeing with يضعوا and يمنعوا) against the copy-text's singular
يصطنع — grammatically the smoother reading, but the copy-text's singular is intelligible as a
shift to the gnomic and the sentence is not one the translation turns on; and witness 2's
أقل شكراً من الإنسان for the copy-text's من الناس, which changes nothing in English.
One loss in witness 2 is a loss and not a variant: في جبل قريب من مدينة appears there without
قريب, leaving the monkey's home on a mountain of a city. Optical, not editorial.
3. What the collation settles for the rendering
Nothing in the sound-figure inventory (R30-v1/translation.md §3) depends on a disputed reading.
All fourteen loci read identically in both witnesses once the recognition noise is discounted, and
the two adopted corrections fall outside every one of them. This is stated because the previous
Arabic work in this project had a figure that existed in one witness and not the other
(../alf-layla/collation.md), and the check was therefore worth making explicitly rather than
assuming.