| Work / author / dates | The Hymn of the Soul, or Hymn of the Pearl (madrāšā d-Yudā Tāʾmā šliḥā), in the Syriac Acts of Judas Thomas; anonymous, probably 3rd century CE, Edessa region |
| Part added | The whole Hymn: 105 couplets in Bevan's numbering |
| Reader route | /read/hymn-of-the-pearl · about and apparatus /read/hymn-of-the-pearl/about |
| Status | working edition, the library's first Syriac text · reviewed by: — |
1. Witnesses
| Siglum | Witness | Kind | Vowelled? | Licence | Location |
|---|---|---|---|---|---|
| MS | London, British Library Add. 14645 (936 CE), the only Syriac manuscript | MS | partly | — | not consulted directly; known through W and B |
| W | W. Wright, Apocryphal Acts of the Apostles (London 1871), vol. 1 pp. 274–279 (Syriac), vol. 2 pp. 238–245 (English) | first edition | partly | public domain | archive.org apocryphalactsa01wriggoog, apocryphalactsa02wriggoog |
| B | A. A. Bevan, The Hymn of the Soul (Texts and Studies V.3, Cambridge 1897), Syriac pp. 10–30, English pp. 11–31, notes pp. 32–40 | critical edition (MS re-examined) | partly | public domain | archive.org hymnofsoulcontai00beva |
| DSC | Digital Syriac Corpus no. 338 ("Hymn of the Soul"), transcription M. Oez, TEI J. E. Walters, status "unedited transcription" | digital | partly | CC BY 4.0 | github.com/srophe/syriac-corpus data/tei/338.xml; copy in scripts/pearl/ |
| G | M. Bonnet, Acta Apostolorum Apocrypha II.2 (Leipzig 1903), Acts of Thomas §§108–113, pp. 219–224 (codex Vallicellianus B 35) | critical edition of the Greek | yes | public domain | archive.org ActaApostolorumApocrypha2 |
Also consulted: DSC no. 332 (the whole Acts of Thomas from Wright, same transcriber), which gives Wright's page breaks.
2. Base and why
DSC 338 is the only openly licensed digital text. It is a transcription and needed checking, so every line was compared with Bevan's printed Syriac (page images of the archive.org scan at 300–600 dpi). The DSC mostly reproduces the manuscript's readings (as Wright printed them), with Bevan's emendation in 43a. This edition keeps that wording and records Bevan's readings as notes, so the reader sees the manuscript and the editor side by side. Numbering: Bevan's 105 couplets, the numbering used by Mead and James.
3. Collation
- Tool:
scripts/pearl/build_pearl.py(maps the DSC's 205 lines onto Bevan's couplets and applies the corrections inFIXES). - Structure: Bevan's lacunae (25b end, 26a, 68a, 71b) are kept as gaps. The DSC had lost the eagle couplet (51) entirely.
- Disagreements between DSC and Bevan's print not explained by Bevan's apparatus: 12, all corrected (section 4). A few further places where the print was hard to read at the resolution available are left as in the DSC and listed in section 12.
4. Decisions (substantive)
| Where | Reading adopted | Rejected (DSC) | Reason | By |
|---|---|---|---|---|
| 13a | ܝܡܐ | ܝܡܢܐ | Bevan's print | AI, checked against the page image |
| 22a | ܢܫܟܒ | ܢܫܟܚܒ | typing error | AI |
| 34b | ܕܥܠܝܗܿ | ܕܥܝܠܗܿ | transposition; cf. 45b, 57b | AI |
| 48a | ܦܨܓܪܝܒܢ | ܦܨܥܪܝܒܢ | Bevan; Wright II p. 241 n. b | AI |
| 51 | ܦܪܚܬ ܒܕܡܘܬ ܢܫܪܐ / ܡܠܟܐ ܕܟܠܗ ܦܪܚܬܐ | (missing) | Bevan p. 20; every translation has it | AI |
| 56a | ܡ̈ܠܟܐ | ܡܗܠܟܐ | Bevan; cf. 33a, 44a | AI |
| 70b | ܕܬܓܪ̈ܐ | ܕܬܓܪ̈ | letter lost | AI |
| 71a | ܕܒܥܒܼܪܗ | ܘܒܥܒܼܪܗ | Bevan and his note | AI |
| 73b | ܐܒ̈ܗܝ | ܐܒܗܗܝ | typing error | AI |
| 74a | ܕܓܙܒܪ̈ܝܗܘܢ | ܕܓܙܒܪܗܝܗܘܢ | typing error | AI |
| 89b | ܚܿܙܝܬܗܿ ܕܡܬܥܬܕܐ | ܚܿܙܝܬܗܿܕ ܡܬܥܬܕܐ | word division | AI |
| 101b | ܐܬܚܠܛܬ | ܐܬܚܾܛܬ | Bevan ('I mingled') | AI |
Bevan's own emendations (26 footnoted places: 9a, 12b, 40b, 43a, 54b, 63a, 65a, 72a, 73a, 76b, 77a–b, 87a, 99b and others) are shown as notes in the word panel and on the About page, not adopted.
5. Quotations verified
None in the Hymn. The bridges cite Quran 27:20–31 (the hoopoe and Solomon's letter), which the Occidental Exile quotes.
6. Vowelling
As in the DSC transcription: partial pointing (seyame, some qushshaya/rukkakha and vowel dots) as printed by Wright/Bevan. No vowels supplied.
7. Segmentation and word IDs
Words split on spaces; proclitics (w-, d-, b-, l-) stay attached, as in Syriac writing. IDs <verse>.<n> (e.g. 51.4 = ܢܫܪܐ 'eagle'), stable as long as syriac.tsv is unchanged. Each word records its half-line (a/b).
8. Analysis layer
No openly licensed morphological resource covers this text: the DSC file has no morphology; SEDRA (Beth Mardutho) covers the Peshitta New Testament and its reuse terms could not be confirmed. Lemma, root, analysis and contextual gloss are therefore by AI (Claude), marked AI in the reader (scripts/pearl/ai/forms.tsv, lexicon.tsv). Payne Smith's Compendious Syriac Dictionary (1903) is linked for every lemma: the page was found with Dukhrana's alphabetical page lookup (one request per lemma, page numbers only); page 299 (margānītā) was checked by eye.
Cognate map. Each lemma may carry an Arabic root with a relation: cognate (inherited, by the regular correspondences š/s, t/th, d/dh, ḥ/ḥ–kh, ʿ/ʿ–gh–ḍ, initial y/w), loan (a borrowing in either direction or a travelling word, with the source named where one is: Bevan's notes, Fraenkel 1886, Jeffery 1938), or ai (a resemblance proposed by AI that breaks the rules or is disputed). Build output (2026-09-28): 306 lemmas; 160 set beside an Arabic root (132 cognate, 12 loan, 16 proposed); 143 of those lemmas (107 distinct Arabic roots) have a root page in the library and are linked. Pronouns and particles (أنا, من, لا …) are shown as Arabic words without a root link. Margānītā ↔ marjān is marked loan (a Wanderwort), not a Semitic root.
9. Commentaries aligned
None. The bridges (app/src/content/pearl/bridges.json) take their place: 9 resonances with Suhrawardī's Qiṣṣat al-ghurba al-gharbiyya (and Ibn Sīnā's ʿAyniyya), one documented later reception (Ernst 2003), two context notes (Plotinus; gnostic, Manichaean or Christian readings). Verdict, from Corbin's own words: typological kinship, not a documented route (research notes with URLs and page numbers in scripts/pearl/research-corbin.json).
10. Translation
Published, public domain, aligned verse by verse: Wright 1871 (prose, divided by the Syriac), Bevan 1897 (same numbering), Mead 1908 (his own numbering, no separate 68), James 1924 (from the Greek, numbered by the Syriac). OCR cleaned by AI (scripts/pearl/tr/*.json, doubts listed in each file). Greek: Bonnet 1903, OCR cleaned by AI (68 corrections), divided by James's numbering; absent from the Greek: 51, 52, 60, 67, 72–74, 85; the Greek has 46 before 45. 30 variant notes (24 from James's remarks, 6 by AI). AI translation: Claude, from the Syriac printed here.
11. Rights and private layers
All public. James 1924: published in the UK in 1924, so its US term ended on 1 January 2020; James died in 1936, so it is also public domain in life+70 countries. Brockelmann's Lexicon Syriacum (1895) was not used: public domain in the US, but Brockelmann died in 1956, so it enters the EU public domain only on 1 January 2027. Downloads are logged in data/raw/pearl/ACQUISITIONS.log; no HathiTrust login was needed.
12. Open questions / to review
- A Syriacist should check the AI lemmatisation, above all the obscure words (27a ܥܝܢܝܢܝ, 35a ܕܛܘܪܗܝܗܘܢ, 66a, 81b ܠܓܘܥܠܢܝ, 90b ܡܝܚܬܝܗ, 103a ܕܪܘܣܐ).
- Lines kept as in the DSC where the print was hard to read at the available resolution: 5b ܐܫܩܝܠܗܿ, 13b ܣܝܩܐ, 27a, 28b ܕܡ̈ܣܝܒܐ (Bevan's English 'the unclean'), 50b ܡܪ̈ܝܪܐ (Bevan may print ܡܪ̈ܝܕܐ 'rebellious'), 61b ܕܐܦܢܐ.
- Wright vol. 1 has no page map on archive.org; the Syriac page links go to the item, not the page.
- Payne Smith pages at the start of a letter may be one page out (Dukhrana's lookup is alphabetical).
- Poirier 1981 (the standard edition of Syriac and Greek) and Ferreira 2002 are in copyright and were not used; worth comparing privately.
13. Change log
- 2026-09-28: added (Syriac, analysis, five published layers, AI translation, bridges, theme-path stop on 'The fall and the return of the soul').