What we document, and what is next.
QuranRecords maps the documented ways the Quran has been counted, written, transmitted, and encoded. This page states exactly what that covers, what is live today, and what is coming - honestly, and cited to source.
Copies of the Quran agree to a degree no other ancient text approaches. What differs is bounded and was catalogued by the tradition's own scholars: how verses are numbered, which marks belong to the original skeleton, and letter-level differences between the canonical readings. We turn those classical records into an open, interactive, cited atlas. This page is our public inventory of it.
The five kinds of documented variation
1. Verse counting - the seven schools (‘adad al-ay)
The same text, divided differently. The schools of Kufa, Basra, Mecca, Medina (two systems), Damascus, and Homs disagree only on where certain verses end - never on the wording - so their totals differ slightly: the Kufan count is 6,236, the Basran about 6,204, the Damascene about 6,226. The differences turn on whether a phrase-break, the opening letters of a surah, or the basmala is counted as its own verse. Documented in classical works such as al-Dani's al-Bayan fi ‘add ay al-Qur’an.
2. The canonical readings (qira’at)
The traditionally transmitted ways of reciting - seven, extended to ten and fourteen - each carried by two named transmitters (rawis). The world's common printed text is Hafs ‘an ‘Asim; across North and West Africa the common reading is Warsh ‘an Nafi’. The readings differ mostly in vocalization and occasionally by a letter; what stays identical beneath them is the larger story.
3. The rasm and the script layers
Beneath the printed page sits the rasm - the bare consonantal skeleton of the Uthmanic text, before the dots and vowel marks that later hands added to guide pronunciation. The regional master-codices sent to the garrison cities carried a small set of documented consonantal differences, catalogued by al-Dani (al-Muqni’) and Abu Dawud. QuranRecords tints each character by its stratum so the seventh-century skeleton is visible inside the modern page.
4. The earliest manuscripts
The oldest witnesses - the Sana’a palimpsest, the Codex Parisino-petropolitanus (BnF Arabe 328), the Birmingham/Mingana leaves, and others - show where an early hand spelled fuller or shorter, and what the archetype carried. Where transcriptions and open image rights permit, we align them to the verse and tie every reading to its folio and its publication.
5. Digital encodings
The layer nobody documents: how electronic editions differ - several Unicode codepoints for one alif, diverging word boundaries, invisible marks - the silent choices inside every Quran app and website, cataloged edition by edition.
What is live today
Counting. Every school's total, the Kufan numbering verified against the text, and a live counter that shows any selection under every documented convention at once.
The Hafs reading, full text, grounded in a verified digital source - the spine the other readings align against.
Script strata and the encodings layer - orthographic layers tinted by age, and documented digital-edition differences.
Translations - 126 editions across dozens of languages: open any verse in any of them, served live and cited to source.
The canonical readings - six riwāyāt (Ḥafṣ, Warsh, Qālūn, al-Dūrī, Qunbul, al-Bazzī): real word-by-word text, differences highlighted against Ḥafṣ, cited to KFGQPC.
Tafsir - 20 works of classical and modern exegesis (Ibn Kathīr, al-Ṭabarī, al-Qurṭubī and more), for any verse, served live.
Per-sūra verse counts, computed from the six riwāyāt. Each reading numbers its own āyāt, so the counts come straight from the text - Ḥafṣ 6236, Warsh & Qālūn 6214, al-Dūrī 6217, Qunbul & al-Bazzī 6220 - with every per-sūra difference highlighted.
Manuscripts - real folios of the earliest witnesses (Birmingham, the Codex Parisino-petropolitanus / BnF Arabe 328, the Blue Qurʾān, the Samarqand codex): openly-licensed images shown inline, each with what to look for and a link to its full record.
What is next
Deeper reading tools - aligning the six readings across their differing verse-numbering, and adding per-riwāya audio recitation.
A manuscript viewer aligning the earliest codices to the verse, starting with the openly-licensed folios.
Open scholarly review - named reviewers for the readings, the counting literature, and the manuscripts, with a public correction log.
Versioned, hashed, citable data releases under a permissive licence, and an open API.
What we still need (no open access yet)
Honesty runs both ways - here is what we cannot yet add, and why. If you can help unlock any of it, tell us.
The canonical readings beyond the six now live. Six riwāyāt are in the atlas; the remaining transmitters of the ten (Ibn ʿĀmir's, Ḥamza's and al-Kisāʾī's, plus Abū Jaʿfar, Yaʿqūb and Khalaf) and the non-canonical readings are not openly digitized as verified text. Stated plainly.
The regional counts the six riwāyāt don't cover - Shāmī and Ḥimṣī - with full line-citation to al-Dānī. The readings we carry already yield the Kūfī, Madanī and Makkī reckonings straight from the text; the Damascus and Ḥimṣ totals exist only in printed critical editions and need careful transcription.
A verse-aligned dataset of the regional rasm differences. No open, machine-readable version exists; it has to be built from al-Dānī's al-Muqniʿ and Abū Dāwūd.
Folio-to-verse alignment, and images for the rights-restricted codices. Openly-licensed folios are now shown; aligning each folio to the exact verse, and clearing reuse rights for the closed collections (e.g. Topkapı, Ṣanʿāʾ), is the step that remains.
How we keep it honest
Every variant names its classical source and carries a provenance grade. Where scholars disagree, we record the disagreement as data - we never resolve it. Numerical-pattern claims are quarantined to a dedicated instrument with hard rules, because our own history made that lesson permanent: this project grew from the 74:30 Project, which tested a 19-based claim and published its retraction after an adversarial audit. The rule that came out of it: don't believe anyone - look.
Help build it
If you study the qira’at, the ‘adad literature, or the rasm; hold manuscript materials or transcriptions; maintain a digital edition; or can read the classical tables - the atlas credits every contribution by name. Write to the team, or join the list for the first datasets and the open call for reviewers.