Theosis Library

About

The Project

Theosis Library is a verified primary source database for ancient texts. The premise is simple: in an age when AI can generate plausible-sounding ancient writing in seconds, the only durable form of authority is the photograph. You can fake a translation. You cannot fake the parchment.

Every chapter in this library meets three non-negotiable criteria:

  1. A real manuscript scan. The actual photograph of the folio of an ancient codex containing the passage. Sourced from public-domain or CC-licensed digitization projects.
  2. The verbatim original-language text. Greek, Hebrew, Arabic, Akkadian, or Old Norse, transcribed from the manuscript itself or from a verified public-domain critical edition. No AI generation. No paraphrase.
  3. The verbatim public-domain English translation. The King James Version (1611) for biblical texts, Brenton (1851) and Lightfoot (1891) for the Septuagint apocrypha and Apostolic Fathers, Pickthall (1930) for the Quran, Campbell Thompson (1928) for Gilgamesh, Butler (1898) for Homer, and Thorpe (1866) for the Edda, all retrieved verbatim from public-domain sources.

Nothing else gets in. If we cannot deliver all three for a chapter, that chapter is not in the library.

Current Holdings

2,362 verified entries across 11 manuscripts, every one passing the three-criteria rule:

  • 260 New Testament chapters from Codex Sinaiticus (4th c., codexsinaiticus.org), with the verbatim Codex Sinaiticus Project transcription of each folio as the primary Greek, Tischendorf's 8th critical edition (1869) beneath it as a comparison reading, and KJV English. Every section links to the specific folio that contains its verses.
  • 240 Old Testament (Septuagint) chapters from Codex Sinaiticus, with the verbatim manuscript transcription preserving uncial forms, lunate sigma, and nomina sacra abbreviations. The OT side excludes books lost from the Sinaiticus codex (Genesis through 1 Chronicles, Hosea, Amos, Micah, Lamentations, most of Esther, parts of Jeremiah, Song of Songs). Psalms are paired with the correct King James psalm across the Septuagint numbering shift, and every chapter where Septuagint and KJV verse divisions differ carries a numbering note.
  • 220 New Testament chapters from Codex Vaticanus (4th c., Vatican Apostolic Library, Vat. gr. 1209), with page-level folio mapping via NTVMR, Tischendorf Greek, and KJV English.
  • 45 Gospel and Acts chapters from Codex Bezae (5th c., Cambridge University Library MS Nn.2.41), the most important Western-text witness, with page-level NTVMR mapping, Tischendorf Greek, and KJV English.
  • 24 books of Homer's Iliad from Venetus A (Marciana Library, Marc. gr. Z. 454, 10th c.), with line-level folio mapping from the Homer Multitext Project, Perseus Greek, and Samuel Butler's 1898 English translation.
  • 11 Poetic Edda poems from Codex Regius (GKS 2365 4to, c. 1270), the unique Icelandic manuscript at the Árni Magnússon Institute, with verbatim Old Norse from heimskringla.no and Benjamin Thorpe's 1866 English translation.
  • 199 further chapters from Codex Sinaiticus: the Septuagint deuterocanon (Tobit in its longer recension, Judith, 1 and 4 Maccabees, Wisdom of Solomon, Sirach), the surviving Song of Songs, and the two early Christian writings bound into the codex after Revelation, the Epistle of Barnabas and the Shepherd of Hermas. English from the King James Apocrypha, Brenton's 1851 Septuagint (4 Maccabees), and J. B. Lightfoot's 1891 Apostolic Fathers.
  • 228 New Testament chapters from Codex Alexandrinus (5th c., British Library Royal MS 1 D. VIII), from the 1909 British Museum photographic facsimile, with NTVMR verse-level page mapping confirmed against the verse caption printed under every plate, Tischendorf Greek, and KJV English.
  • 88 Gospel chapters from Codex Washingtonianus (4th-5th c., Freer Gallery of Art, F1906.274), from the Smithsonian's own image service, with CSNTM verse-to-page lookup, Tischendorf Greek, and KJV English. The Freer Logion after Mark 16:14 is pointed out on its page.
  • 929 chapters of the Hebrew Old Testament from the Leningrad Codex (1008 AD, National Library of Russia, Firkovich B19a), the oldest complete Hebrew Bible, with the folio-by-folio verse index of tanach.us cross-checked against Sefaria for every chapter, the Unicode Leningrad Codex transcription, and KJV English.
  • All 66 chapters of Isaiah from the Great Isaiah Scroll (1QIsa-a) (c. 125 BCE, Israel Museum), cropped column by column from the public-domain composite photograph, with the scroll's own transcription (ETCBC Dead Sea Scrolls dataset) rather than the medieval text, and KJV English.
  • 33 suras of the Quran from the Codex Parisino-petropolitanus (BnF Arabe 328, 7th-8th c., one of the oldest Quran manuscripts), each folio side mapped by Corpus Coranicum, with the Tanzil Uthmani text and Pickthall's 1930 English.
  • 19 Nineveh tablets of the Epic of Gilgamesh (7th c. BCE, British Museum: Tablets I, VI and XI including the Flood Tablet K.3375), with the Electronic Babylonian Library's tablet-by-tablet Akkadian transliteration and R. Campbell Thompson's 1928 English, line-concorded.

Why This Matters

AI can generate plausible translations of any ancient text in seconds. What AI cannot generate is a photograph of a manuscript. The scans in this library are the evidence. The original-language texts are verifiable against those scans. The translations are checkable against the originals. Every claim links to a source you can see with your own eyes.

As AI becomes more capable, the need for a verification layer becomes greater, not less. This is that layer.

Contact

Corrections, alternate translations, and scholarly correspondence are welcome at contact@theosislibrary.com.