| Work / author / dates | Kitāb al-sīra al-falsafiyya (كتاب السيرة الفلسفية), a short polemical-autobiographical treatise in prose · Abū Bakr Muḥammad b. Zakariyyā al-Rāzī (Rhazes), physician-philosopher of Rayy, c. 250–313 AH / c. 865–925 CE (dates vary: c. 865–c. 925 in the British Library catalogue; the OpenITI tag is 313). The treatise carries no date; it speaks of the author's failing sight and impaired hand after fifteen years on the Jāmiʿ and of "nearly two hundred" works (¶11, ¶10), so it comes late in his life (Adamson's reading is that it is an old-age text and a defence against a charge of hypocrisy: an inference, review queue G1). English title in the reader: «The Philosopher's Way of Life». |
| Part added | The whole treatise, Kraus 1939 pp. 99–111 (pp. 97–98 are Kraus's introduction, not reproduced): 12 reading sections, 3,003 words (3,017 text tokens with the dashes), 479 distinct roots. |
| Reader route | /read/razi-sira · apparatus /read/razi-sira/about · public since 13 Oct 2026 (§11). |
| Status | published 13 Oct 2026 (on the editor's explicit instruction), working edition · Arabic: Kraus 1939 read from a page scan by AI (three sonnet readings, settled against OpenITI and by an opus adjudication of the disputed places), not yet checked by a person · one printed page (p. 105, 241 words) and a 24-word line that OpenITI omits are restored from Kraus's pages (§3; review queue B) · vowelling = full AI vocalisation, Kraus's printed marks kept as his (§6) · analysis by CAMeL with 1,109 of 3,003 words corrected by an AI sweep · English (readable and literal) and word glosses by AI · reviewed by: — (published without a person's review; no person has checked the Arabic, the vowelling, the roots or the English; an AI (opus) review of 13 Oct 2026 settled the review queue, razi-sira-review.md) |
What the treatise says. Some critics, "people of inquiry and discernment", say the author, by mixing with people and earning his living, has abandoned the way of life of the philosophers and above all of "our leader Socrates" (¶1). He grants the facts about Socrates (no kings, no meat, no children, a jar in the wilderness) but says they belong to the beginning of his course; later Socrates fathered children, fought, went to gatherings and ate well except meat, and the difference between him and us is of degree, not kind (¶2). On principles taken from his other books (afterlife by conduct; knowledge and justice as the aim of life; reason against desire; a merciful Lord; no pain taken for a lesser pleasure) he derives two corollaries: no pain for a pleasure that the pain outweighs (¶3–4), and no needless pain to any sentient being, with the cases of work animals, the two travellers with one skin of water, and the killing of predators (¶5–6); and so none to oneself: the Indians, Manichaeans, monks and Muslims who torment themselves are in the wrong (¶6). People differ by upbringing, so the philosophic life has an upper limit (no injustice, no killing) and a lower one (frugality without torment); between them all is permitted; beyond the lower limit one leaves philosophy (¶7–8). The sum: God is knowing and just; the servant closest to Him is the most knowing, just and merciful; "philosophy is likeness to God as far as human capacity allows" (¶9). The defence is in two parts, as philosophy has two parts: knowledge (a list of his books, "some two hundred") and action (physician and table-companion to the ruler, not a soldier or administrator; twenty thousand leaves in a year, fifteen years on the Jāmiʿ, sight lost) (¶10–11); and a challenge: show me my fault in knowledge, or else benefit from my knowledge and ignore my conduct (¶12).
| ¶ | Section | Kraus pp. | Words | OpenITI lines |
|---|---|---|---|---|
| 1 | The charge: a courtier who departs from Socrates | 99 | 156 | 41–53 |
| 2 | Socrates early and late: a difference of degree, not of kind | 99–101 | 401 | 53–85 |
| 3 | The principles on which the philosophic life rests | 101–102 | 232 | 86–106 |
| 4 | First corollary: no pain bought for a lesser pleasure | 102–103 | 364 | 106–134 |
| 5 | Second corollary: no needless pain to any sentient being | 103–105 | 391 | 135–159 |
| 6 | Other animals; and no pain to oneself: Indians, Manichaeans, monks | 105–106 | 246 | 161–166 |
| 7 | Different upbringings; the upper limit | 106–107 | 194 | 167–182 |
| 8 | The lower limit, the middle that is permitted, and what lies beyond | 107–108 | 219 | 183–201 |
| 9 | The sum: God knowing and just, and philosophy as likeness to God | 108 | 108 | 201–209 |
| 10 | The defence, first part: knowledge | 108–109 | 306 | 210–232 |
| 11 | The defence, second part: conduct | 109–110 | 255 | 232–250 |
| 12 | The challenge to the critics, and the close | 110–111 | 145 | 250–263 |
0. The title, and what the edition says about the manuscript
Kraus (pp. 97–98): the title appears as «كتاب السيرة الفلسفية» in al-Bīrūnī's list and in the manuscript; Ibn al-Qifṭī and Ibn al-Nadīm give «كتاب (في) السيرة الفاضلة»; Ibn Abī Uṣaybiʿa has three titles that probably all point here («كتابا في سيرة الحكماء», «كتابا في سيرته», «كتابا في السيرة الفاضلة وسيرة أهل المدينة الفاضلة»). The text reached us in one collection, London, British Library Add MS 7473 (ff. 1v–5v), copied in 639 AH (1242 CE) in a fine naskh hand "with not a few mistakes"; Kraus corrected in the text and gave the manuscript's readings in the footnotes under the siglum خ. The manuscript's title page names the author «الحكيم أبو نصر (كذا) محمد بن زكريا الرازي المتطبب». Kraus had published the work in Orientalia n.s. 4 (1935) with an introduction and French translation, and prints it here "again with more care in the correction". Whether other witnesses exist is not established here (the apparatus has only the siglum خ in the AI reading).
1. Witnesses
| Siglum | Witness | Kind | Vowelled? | Licence | Location |
|---|---|---|---|---|---|
| K | Paul Kraus (ed.), Rasāʾil falsafiyya Part 1, Cairo: Fuʾād I University, Faculty of Arts 22, Maṭbaʿat Bawl Bārbīh, 1939; the Sīra at pp. 99–111 | standard (critical) edition, from the one manuscript | sparsely (light marks printed) | likely public domain, not legally cleared (§11) | read from archive.org elshandawily10263 (a private upload, 331 pages, title page of the 1939 Cairo print; whether first printing or facsimile reprint is not established); preferred link for readers NYU Arabic Collections Online (from the American University in Cairo's copy, B753 .R38 1939), https://sites.dlib.nyu.edu/viewer/node/24761, hdl 2333.1/8gthtdkh, high-resolution PDF and IIIF (336 canvases; printed p. n = canvas n+12) |
| خ | London, British Library, Add MS 7473, ff. 1v–5v, copied 639/1242 (colophon on f. 5v) | the one manuscript | — | public domain as a work; image licence not read | QDL https://www.qdl.qa/en/archive/81055/vdc_100023601232.0x000001 (the site refused scripted access, 13 Oct 2026); known only through Kraus's footnotes |
| O | OpenITI 0313IbnZakariyyaRazi.SiraFalsafiyya.Shamela0012834BK2-ara1, the typed text of the Beirut 1982 reprint of Kraus (Lajnat iḥyāʾ al-turāth al-ʿarabī, Dār al-Āfāq al-Jadīda, 5th ed.), via al-Maktaba al-Shāmila book 12834 | digital, typed (status pri) | no | CC BY-NC-SA 4.0 (OpenITI corpus) | OpenITI/RELEASE commit cfc4157a; committed whole at data/sources/openiti/ (SHA-256 6ffd041d850ea3aba9bda8bccfc0535685a896942aa21a20e327f53f48cf9f19); every section links its lines |
| K35 | Kraus, "Raziana I: La conduite du philosophe", Orientalia n.s. 4 (Rome 1935), pp. 300–334 | first edition with French translation | — | UNVERIFIED | not seen; cited |
| — | Frankfurt reprint of K, Institut für Geschichte der Arabisch-Islamischen Wissenschaften, 1999 (a private upload, 20230124_20230124_0254) | facsimile reprint | — | new front matter in copyright | not used (private reference only; nothing was copied from it) |
| — | McGinnis and Reisman, Classical Arabic Philosophy (Hackett 2007), ch. 2; Arberry, "Apologia pro vita sua" (1967, UNVERIFIED) and the passages in The Spiritual Physick of Rhazes (1950); Adamson, Al-Rāzī (OUP 2021) | translations and study | — | in copyright | cited only; none of their wording is used; not used for triangulation |
2. Base and why
K, read from its page images. The editor's instruction was that Kraus 1939 is the base. It is the standard edition; the one manuscript is known only through it; and OpenITI (O) is a typed copy of a 1982 reprint that has lost a printed page (§3) and about forty slips. O is therefore a witness and the digital starting point, never the base. The scan used for reading was found by searching archive.org on 13 Oct 2026 (a private upload; the NYU scan found later is the better public link and carries a library's public-domain belief statement). The first AI reading (three sonnet subagents, five pages each) was compared with O word by word; the 72 places of disagreement were then looked at on the page images by an opus subagent, which settled 24 questions and corrected eight of the first reading's words (listed in razi-sira-base.json → changes, type reading). The adjudicator also showed that the first reader had, in places, copied a manuscript variant from Kraus's footnote into the text (the case of «ونقتضبها اقتضابا»: the footnote's «ونقتضيها اقتضاء خ» is the manuscript's reading, the text has the other), so a second look at any place where a footnote exists is advisable (review queue C2).
3. Collation (scripts/razi_sira/collate.py → data/sources/razi-sira-collation.json)
- Letters only (marks, hamza seats, ى/ي, ة/ه normalised). Base 3,002 words, O 2,719.
- 51 differing spans (48 after the AI review of 13 Oct 2026, which found the print agreeing with O at «يدعواننا» and «سننهم»): 13 where K has words O lacks, 38 where they differ in a word (OpenITI slips such as «يجوز» for «يجور», «أوردنا» for «أردنا», «بالحم» for «باللحم», «والسراب» for «والشراب», «قي» for «في», «الجدين … جددت» for «الحدين … حددت», «أسرف» for «وسرف», word-division and spelling differences such as «ههنا»/«هاهنا», and a few readings the adjudication did not look at, review queue A9).
- Two omissions of O matter. (1) A whole printed page, Kraus p. 105: O jumps from «أتلفت حيوانات كثيرة،» (end of p. 104, line 19) straight to «من الترهب والتخلي» (the first words of p. 106; the end of p. 105's last line «… ما يستعمله النصارى» is lost with it). The restored passage is 241 words: from «هذا باب خاصّ بأمثال هذه الحيوانات» (the second reason for killing predators: a soul can be freed only from a human body), through the snakes and scorpions, the animals that are worked, the philosophers' disagreement about meat, to the Indians, the Manichaeans and the Christians. The scoping note guessed about 130 words at pp. 103/104 from the Frankfurt OCR; the real gap is this page, and O's own page tags put its resumption at "P104" (O has no P105 tag). (2) 24 words at p. 109 (the middle of the list of the author's works: «الزمان والمكان والمدة والدهر والخلاء» … «سبب تحرك الفلك على استدارة»).
- Source of the restoration: Kraus's own pages (the archive.org scan of the 1939 print), not the 1999 Frankfurt reprint (not used: FLAGS has the note that nothing was copied from it). Read twice by sonnet and once by the editor-AI; no person has read it against the page (review queue B1–B2).
- The page tags of O are used only for the links; Kraus's page numbers are the anchors (section labels say "Kraus pp. …"; the exact word at which each page begins is in
razi-sira-base.json→ passages[].pages).
4. Decisions (substantive)
| Where | Reading adopted | Rejected | Reason | By |
|---|---|---|---|---|
| p. 99 | «محمد بن زكريّاء الرازىّ» | O's «الرازي» alone | printed (adjudication 1) | AI (opus) |
| p. 100 | «وحبِّه» | «وحبٍّ» (first reading) | print has ه | AI (opus) |
| p. 101 | «ونقتضبها اقتضاباً» | «ونقتضيها اقتضاءً» | print has the one-dot ب; the ي-form is the manuscript's (footnote 13) | AI (opus) |
| p. 104 | «يُجار عنه», «البَشِع», «الإعناف», «العيشة» | «يجارعنه», «البشيع», «الأعنف», «العيش» | AI (opus) | |
| p. 106 | «إصابةَ لذّةٍ مّا» | «اللذّةِ» | print is indefinite | AI (opus) |
| p. 108 | «سُنَنهم» (queue A3) | «سنتهم» (first reading, adjudication, editor-AI) | thresholded full-resolution NYU crop: two dots staggered over two teeth; agrees with O and with the plural «بسيَرهم»; probable | AI (opus) review 13 Oct, open for a person |
| p. 99 | «ولا يُنسِل» | «ولا ينْسِل» | the print marks ḍamma and kasra (form IV) | AI (opus) review 13 Oct |
| p. 101 | «يدعواننا» | «يدعوانا» | the print has two nūns (the first reading dropped one); now agrees with O | AI (opus) review 13 Oct |
| p. 105 | «فهذا باب خاصّ», «وممّا يدخل» | «هذا», «ومّا» | NYU scan, the restored page read line by line | AI (opus) review 13 Oct |
| p. 109 | «المُفْنُون لأعمارهم» | «المُفنْونُ» | marks as printed | AI (opus) |
| p. 111 | «ونُذعِنَ», «فَإِنْ» | «ونُدعِنَ», «فَاِنْ» | AI (opus) | |
| spacing | «ولا ينْسِل» | the print runs «ولاينْسِل» together | word division restored (a changes entry, type mechanical) | editor-AI |
| Kraus's supplements | five angle-bracket words printed without brackets | listed in supplied (queue C1) | editor-AI | |
| paragraphs and sections | 12 sections | our own (§7) | editor-AI |
5. Quotations verified
The treatise quotes no āya. scripts/quran_refs.py's automatic skeleton match returned one hit, the opening basmala plus «قال» matching 72:1 («بسم الله الرحمن الرحيم قل …»), a false positive (the basmala is a formula); the text is therefore not registered in scripts/quran_refs.py, so it adds nothing to the public Quran index. Other quotations: the proverb «لكل جديد لذة» (¶2); the saying of the philosophers «الفلسفة هي التشبه بالله عز وجل بقدر ما في طاقة الإنسان» (¶9); the book titles; one unattributed verse (¶12, «اعمل بعلمي فإن قصرت في عملي // ينفعك علمي ولا يضررك تقصيري»; metre and author not traced).
6. Vowelling
- Full AI vocalisation (13 Oct 2026, the editor's instruction). Every one of the 3,003 words is vowelled in full, case endings and verb moods included (no pausal forms), by Claude (Sonnet 5.5) writing each section over the earlier light vowelling (
data/sources/razi-sira/vowelling-full/N.txt, validated and merged byscripts/razi_sira/full_vowel.py: letters identical to the base; every base mark must survive).vowelMask:t= a letter or a mark of Kraus's print (736 marks, 566 words),a= a mark added by the AI (9,958). The print's marks were themselves read from the page images by AI (§4), so "Kraus's" means the base's marks. Roots, lemmas, ids and the word glosses are untouched (the text and mask of each word are replaced inrazi-sira-analysis.json). - Doubts: 78 places with a defensible alternative (grammatical alternatives, rare forms, proper names, printed marks that look like slips) are in
razi-sira-review.md("Full vowelling: doubts") and in the word panel's two-readings notes (vowelNotes). Check first: ¶6 «أُطْلِقَ», the print's ḍamma kept against the earlier opus decision (active verb). (Settled by the vowel review below: active.) - CAMeL:
scripts/razi_sira/full_vowel_conflicts.pyre-ran only the vowel comparison (the best-agreeing CAMeL analysis of the word's own lemma): 92 words conflict (was 68 against the light marks); they are drawn as doubtful and listed in the review file. - Conventions: shadda on sun letters after the article; tanwin fatḥ on the alif as Kraus prints; a bare alif where the print has no hamza gets a kasra/fatḥa/ḍamma, never a hamza; dual constructs end in yā' with sukūn.
- AI (opus) vowel review, 13 Oct 2026 (an AI check, not a person's review;
razi-sira-review.md→ "Full vowelling: AI (opus) vowel review"; FLAGS O). All 78 doubts settled: 63 as written, 6 changed (one of them, ¶3 «الْجَسَدَانِيَّةِ», still open), 9 more left open as word-panel notes, plus one new note (¶4 «رُطَبِ»). The ¶6 conflict is settled on the NYU crop of p. 105 l. 12: the sign over the alif of «أطلق» is the hamza, with no vowel, so the base's ḍamma was a misreading; the verb is active, «لَمَا أَطْلَقَ حُكْمُ الْعَقْلِ ذَبْحَهَا» (the English already said so). Two other printed marks re-read on the scan: ¶5 «كالأُسُد» (ḍamma on the hamza, plural of asad), ¶7 «غَذِيُّ نِعْمَةٍ … غَذِيُّ بُؤْسٍ» (only a shadda is printed). These three base corrections are inbuild_base.py(FIX). Other vowel fixes: ¶2 «عُنِىَ», ¶3 «الْجَسَدَانِيَّةِ», ¶4 «الْفُيُوجَ … وَالْجُنْدَ أَجْرَأَ» (objects of نرى; the print marks nothing there), ¶5 «أَصْلَحَ» (predicate of كان), ¶8 «الرَّثَّةِ», ¶10 «الْكِيَانِ», «فَقُرَابَةُ». The filler (autofillinfull_vowel.py) no longer puts a fatḥa on the plural wāw before the silent alif (15 words) or on the lām of Kraus's «إلاَّ / لئلاَّ» (11 words). CAMeL: 3 conflicts fixed by those changes, 84 cleared as right (vowelling-full/camel-cleared.json, read byfull_vowel_conflicts.py), 5 still drawn as doubtful. Spot check of ¶4 and ¶9 word by word: 3 wrong endings in ¶4 (360 words), none in ¶9 (108 words), all three in one clause: 3 of 468 words, 0.6%. Kraus's marks are now 731, AI marks 9,941. - Comparison with authoritative vowelled editions: none exists. Not reviewed by a person.
- (Earlier, same day: a light AI vowelling, 957 words marked; superseded.)
7. Segmentation and word IDs
12 sections following the treatise's own turns (the «فنقول» openings and the short last lines of paragraphs on the page; section 5 gathers the second corollary and the beasts, section 6 the other animals and self-torment; 9 is the sum; 10–12 the defence). Each section records its Kraus pages and the word at which each page begins, the manuscript's folio breaks as Kraus marks them (2r, 2v, 3r, 3v, 4r, 4v, 5r, 5v; his «|» and the margin note ٢و …; the last is printed «ظ» only), and its OpenITI lines with a GitHub link. Word IDs rs:<section>:<word>, punctuation tokens rs:<section>:p<n>. Printed paragraph breaks are not marked in the AI reading except where the page shows them by a short last line; the sections are ours.
8. Analysis layer
CAMeL Tools 1.6 (calima-msa-r13 + MLE) via scripts/fusus/analyze_ch.py razi-sira.json razi-sira-analysis.json rs, constrained by the marks: 3,003 words, all analysed, 1,849 with a root, 479 distinct roots, 68 vowel conflicts. Corrected by an AI sweep (sonnet, 4 subagents, each word read in its sentence with the AI English; scripts/razi_sira/apply_fixes.py, fixes in data/sources/razi-sira/sweep/): 1,109 of the 3,003 words changed: 223 roots (kept as rootFix: {from, to, by}), 1,060 lemma / part of speech / features / clitic split / gloss changes (kept as lemmaFix: {from, by, reason}). The sweep found the same kinds of error as the Ibn al-Fāriḍ one: an read as anna; inna/anna/in not tagged as particles or the conditional in read as inna; verbs read as nouns and nouns as verbs; names and the Manichaeans (المنانية) mistagged; case and person/number wrong; clitics wrongly cut; wa- glossed "while" (set to "and" by a script pass). Thirteen cases it could not settle are in the review queue (F). No independent sample check of the sweep's accuracy was run at build (22 changed words were looked at by the editor-AI and all were plausible; the Ibn al-Fāriḍ sweep's own check was 15/15). AI (opus) review, 13 Oct 2026: a random 30 of the 1,109 changes read in context: 0 wrong, 2 incomplete (fixed); 129 stale stem-segment parts of speech synced; the 13 doubts settled (data/sources/razi-sira/lemma-review/, apply_fixes.py … sweep … lemma-review). Not a person's review.
9. Commentaries aligned
None exists. Ibn Abī Uṣaybiʿa and al-Bīrūnī list the work; no classical commentary is known to us. Not claimed absent in the world.
10. Translation
AI English, readable and literal for each section, a contextual gloss for every word and term notes: Claude sonnet, four subagents (three sections each), told to translate from the Arabic with the glossary in data/sources/razi-sira-meta.json (sīra "way of life", al-tashabbuh bi-llāh "likeness to God" (changed to "becoming like God" in the AI review), ṭibb rūḥānī "Spiritual Medicine", mutafalsif "would-be philosopher", ḥadd min fawq / min asfal "upper limit / lower limit", al-mālik "the Owner", al-bārī "the Creator" …) and not to consult or imitate McGinnis–Reisman, Arberry or Adamson (none was supplied; I read the finished English once for sense and found nothing to change but a missing dot in «Muḥammad» in ¶1, fixed in data/sources/razi-sira/review/translation-fixes.json). Each model also listed its own doubts: 121, queued (review queue E). All ✦ AI, unreviewed.
- AI (opus) review, 13 Oct 2026: every one of the 121 doubts settled against the Arabic; 31 sentence revisions in our own wording (log:
data/sources/razi-sira/review/translation-revisions-log.json; applied throughtranslation-fixes.jsonbyassemble_ai.py); the whole English read once for accuracy and register (divine pronouns capitalised; «المتفلسف» "one who practises philosophy"; «التشبّه» "becoming like God"; «المالك» "Owner" throughout). No published translation consulted. Still AI, unreviewed by a person.
10b. Translation triangulation (PROTOCOLS §17)
- Held: no public-domain English translation. McGinnis–Reisman 2007, Arberry and Adamson 2021 are cited only on the editor's instruction (copies exist privately; they were not opened for this build).
scripts/triangulate.py razi-sirawas not run. No "Translators differ" layer exists. If it is run later (private layer), the English may only be checked for contradictions and any contradiction is a bug to fix in our own words.
11. Rights and private layers
- Kraus 1939 (K): "likely, not legally cleared". Reasoning: Kraus died 14 Sep 1944 (TDV İslâm Ansiklopedisi, via search snippet), so life+70 expired 31 Dec 2014 in the EU and most Berne countries. For the US: a 1939 foreign work is protected for 95 years from publication (to 2034) only if its copyright was restored by the URAA on 1 Jan 1996, which needs it to have been protected in the source country (Egypt) on that date; Egypt's term then was life+50 (general knowledge, UNVERIFIED), which for a death in 1944 ended in 1994, so the work would probably not have been restored. Not a legal opinion; not cleared by any adviser. Corroboration, not proof: NYU's page for its scan of the same volume says "NYU believes the materials displayed on this site are all in the public domain" (with a takedown notice), read 13 Oct 2026; and the Frankfurt reprint's new front matter is a separate matter and is not used.
- OpenITI (O): CC BY-NC-SA 4.0 (corpus licence,
docs/SOURCES.mdrow 12). Committed whole with its header and hash. It is a witness; the base is read from K's images, so no OpenITI wording enters the base except where the two agree on Kraus's text. Derivatives of O would be share-alike; the base text does not derive from it by copying. - Page images of K are not rehosted; readers are sent to the NYU scan. The AI reading of K (
data/sources/razi-sira/kraus-reading/) is committed as our own working transcription of a likely-public-domain print. - Translations (McGinnis–Reisman, Arberry, Adamson): cited only, never copied, none in the private layer either.
- AI layers: the vowelling, roots/lemma corrections, English, glosses and notes are AI, unreviewed, labelled in the reader and in the versions record (
app/src/content/versions/razi-sira.json). - Published 13 Oct 2026 on the editor's explicit instruction, modelled on
ibn-al-farid-hubb: id removed fromHIDDEN_TEXTS; the library and catalogue entries moved intoLIBRARY/CATALOGUE(kind: philosophy); word detail moved fromapp/src/data/hidden/words/razi-sira/toapp/public/data/words/razi-sira/; registered inscripts/build_data.py(EXTRA_TEXTS: 58 root pages touched or added, dictionary entries only); search index and connections file rebuilt. Not registered inscripts/quran_refs.py(no āya is quoted; the basmala/72:1 match is a false positive and is excluded). The reader and About notes stay: vowels, English, roots and glosses AI and not reviewed by a person; source note (Kraus 1939 pp. 99–111, read from page scans by AI, checked against OpenITI; OpenITI's missing p. 105 and p. 109 line restored from Kraus's pages); rights "likely public domain, not legally cleared".
12. Open questions / to review
All Arabic judgement calls are queued in docs/texts/razi-sira-review.md (A readings, B restored passages, C Kraus's supplements and apparatus, D contested terms, E 121 translation doubts, F analysis doubts, G secondary claims). In order of weight: B1 (the restored page, read against the NYU scan by a person), A2/A3/A6/A7 (four doubtful print readings), the whole apparatus (C2), the English for the places the model doubted (E), the independent check of the sweep (F-standing). Also: a person's check of the vowelling, the roots and the English; the legal question in §11 (a rights adviser) before publication; collation against the manuscript images on the QDL once their licence is read; triangulation if a copy of McGinnis–Reisman may be used privately; the companion sweep's QDL item is unverified.
13. Connections to the site's texts (editor's proposals; none is a verified lineage)
ikhwan(Brethren of Purity, Epistle 13 close): the same definition of philosophy as likeness to God as far as human capacity allows (al-tashabbuh bi-l-ilāh bi-ḥasab al-ṭāqa al-insāniyya,docs/texts/ikhwan.md§14 has its lineage: Theaetetus 176b through the Alexandrian prolegomena). Al-Rāzī's formula is the earlier, early-tenth-century witness; strongest link. Not stated as influence.- Plato on the site (
alcibiades-1-mirror,republic-sun-line-cave,phaedrus-wings,symposium-diotima): the reception of the Socratic life; al-Rāzī's Socrates is the Arabic tradition's, who fathered children and went to war. No Theaetetus, Apology or Phaedo is on the site. A resonances pass (PROTOCOLS §16; paid run on the owner's machine) has not been run: it would add the web-verified reference checks. hayy-tufaylandhayy(Ibn Sīnā): the visionary route to the same end, against al-Rāzī's rationalist ethic; a parallel, not a lineage.- The Mukhtār al-ḥikam (al-Mubashshir) Socrates chapter is in the scoping notes (
gnomologia-mubashshir-siwan.md), not yet on the site. ibn-sina-ishq,maqalid: vocabulary hits only; unexamined.
14. Companions (PROTOCOLS §18; app/src/content/companions/razi-sira.json)
Link-out only, checked 13 Oct 2026 by fetching each page: the Stanford Encyclopedia entry on al-Rāzī (Adamson); History of Philosophy without any gaps, ep. 126 (Adamson on al-Rāzī; about the five eternals, the Sīra is named in a comment reply); the Abbasid History Podcast ep. 32 (unverified: description does not mention the Sīra); NYU's digitised Kraus 1939 volume 1; the Wellcome catalogue record of the Frankfurt reprint; Hackett's page for McGinnis–Reisman; Wellcome's record of Arberry's Spiritual Physick; the QDL manuscript page (unverified, 403). No recording or reading of the Arabic text was found. Dropped as unverifiable: Encyclopaedia Iranica (403), IEP (404), EI2/EI3, the Orientalia 1935 catalogue page (410). No lecture on the Sīra itself was found.
15. Checks (13 Oct 2026, per PROTOCOLS)
- ESLint on the changed files: clean.
node scripts/design_audit.mjs --lint: 0 off-system values. - Public build (no
PRIVATE_LAYERS) with the text absent:next build --webpackpasses (the default Turbopack build refuses this worktree's symlinkednode_modules; not a code problem)./read/razi-sirais not generated (no entry under.next/server/app/read), the sitemap has no mention, and the hidden id is absent fromapp/public/and from the client chunks (.next/static); it appears only in server chunks and file-trace lists, by design (hiddenTexts.tsis server only). - Private build (
razi-private,PRIVATE_LAYERS=1, port 3130):/read/razi-sira,/about,/sheet, the library entry and/data/hidden/words/razi-sira/p1.jsonanswer 200;/read/ibn-al-farid-hubbunchanged. - Screenshots of the private page at 1280 and 375 px: scratchpad
razi/shots/razi-1280.png,razi-375.png(not committed).
16. Change log
- 2026-10-13: record created. Scan found (archive.org
elshandawily10263; NYUauc_aco000121), read by AI from the page images; compared with OpenITI; two omissions of OpenITI restored from Kraus's pages (p. 105, 241 words; p. 109, 24 words); 24 questions adjudicated by opus; base built (scripts/razi_sira/build_base.py), 12 sections; light AI vowelling; CAMeL analysis; AI English (readable and literal) with glosses and 121 recorded doubts; sonnet lemma sweep (1,109 words changed); versions record, companions, review queue; wired as a hidden text (HIDDEN_TEXTS,PRIVATE_LAYERS=1), launch configrazi-private(port 3130). - 2026-10-13: AI (opus) review, 13 Oct 2026 (an AI review, not a person's; no reviewer is named in
reviews/). Read on the NYU full-resolution scan: queue A settled except A3 (changed to «سُنَنهم», probable) and the unread part of A10; B1 and B2 read line by line (two corrections on p. 105: «فهذا», «وممّا»); base changes «يُنسِل», «يدعواننا»; five apparatus readings corrected; Kraus's supplements confirmed; terms decided (D1 "becoming like God", D3 "one who practises philosophy", D6 "Owner"); six AI vowel marks corrected; 13 lemma doubts settled; sweep spot check 30/30 without a wrong correction (2 incomplete, fixed; 129 segment parts of speech synced); 121 translation doubts settled, 31 revisions. Rebuild:build_base.py→assemble.py data/sources/razi-sira/vowelling→ CAMeL (analyze_ch.py, only the words whose text changed were taken over) →apply_fixes.py data/sources/razi-sira/sweep data/sources/razi-sira/lemma-review→collate.py→assemble_ai.py→reader_data.py. Still hidden; not registered. - 2026-10-13: PUBLISHED on the editor's explicit instruction (PR #52). Before publication: Kraus's five angle-bracket supplements (¶1 «عنه», ¶2 «يَفْضُلُ», ¶3 «و», ¶10 «ما», ¶11 «أنّى لم أَتَعَدَّ») are marked in the reader as editorial (angle brackets, muted, tooltip "Added by the editor (Kraus 1939)";
editorialon the words,scripts/razi_sira/mark_words.py), and the reading A3 «سُنَنهم» (p. 108) is marked as a probable reading (dashed amber underline as for uncertain readings;probable), still open in FLAGS for a person. Unhidden as in §11; reviewed by: — stays. - 2026-10-13: Full AI vowelling of all 3,003 words (§6; Kraus's 736 printed marks kept as
t, 9,958 AI marksa; 78 doubts and 92 CAMeL conflicts for the opus review; FLAGS N). The paragraph-level literal English line is no longer displayed in the reader (library-wide change; the word-by-word glosses remain). - 2026-10-13: AI (opus) vowel review, 13 Oct 2026 (an AI check, not a person's; §6, FLAGS O): the 78 doubts and the 92 CAMeL conflicts settled; ¶6 «أَطْلَقَ» active on the print; three printed marks re-read on the NYU scan and corrected in the base; 13 words revowelled, 26 filler slips removed; four word glosses brought into line (¶5, ¶7, ¶10).