Reference · compiled 31 August 2026
There is no single number of letters, words or verses — there are layers of declared method, and each layer has named authorities who disagreed. This is the list, the rules, and what each rule is worth, measured.
Four layers of choice stack on top of one another. Pick differently at any layer and every number changes.
Before anything is counted, the text itself must be fixed. Ten canonical readings, each with two transmitters, gives twenty canonical forms.
| Transmission | Where it is read | Verse system it uses |
|---|---|---|
| Ḥafṣ ʿan ʿĀṣim | Most of the Muslim world — the Cairo 1924 lineage | Kūfan |
| Warsh ʿan Nāfiʿ | Morocco, Algeria, Mauritania, much of West Africa | Medinan |
| Qālūn ʿan Nāfiʿ | Libya, Tunisia, parts of Qatar | Medinan |
| al-Dūrī ʿan Abī ʿAmr | Sudan, Horn of Africa, parts of West Africa | Baṣran |
| + 16 further transmissions | Preserved and taught; rarely printed for general use | various |
| Rasm | Al-Fātiḥa 1:2 | Letters |
|---|---|---|
| Uthmānī — the printed muṣḥaf | ٱلْحَمْدُ لِلَّهِ رَبِّ ٱلْعَٰلَمِينَ | 17 |
| Imlāʾī — modern simple spelling | الحمد لله رب العالمين | 18 |
84% of āyāt have a different letter count under the two orthographies. Across the whole text the gap is 5,323 letters. Most published counts never say which was used.
Seven regional systems, each traced to early authorities in a specific city, each documented sūrah by sūrah. The primary classical reference is Abū ʿAmr al-Dānī (d. 444 AH), al-Bayān fī ʿadd āy al-Qurʾān.
| System | Total āyāt | Notes |
|---|---|---|
| Kūfa | 6,236 | Transmitted from ʿAlī b. Abī Ṭālib; the printed standard since 1924 |
| Damascus (Shām) | ~6,226 | |
| Medina I | ~6,217–6,221 | Used with Warsh and Qālūn |
| Medina II | ~6,214–6,219 | |
| Mecca | ~6,217–6,219 | Counts the basmala, like Kūfa |
| Ḥimṣ | — | Recorded in the classical lists; rarely printed |
| Baṣra | ~6,204–6,205 |
| Rule | What varies |
|---|---|
| The basmala | Kūfa and Mecca count it as āyah 1 of al-Fātiḥa. Medina, Baṣra and Shām do not — and instead split the final āyah in two, so al-Fātiḥa is seven either way. |
| Disjoined letters الٓمٓ | In the Kūfan system the ḥurūf muqaṭṭaʿa stand as independent āyāt in certain sūrahs. Other systems attach them to what follows. |
| Verse-break placement | The largest source of difference, and it runs sūrah by sūrah. Al-Baqara is counted 285 (Medina/Mecca/Shām), 286 (Kūfa), or 287 (Baṣra). |
Each rule below is a genuine position held by some counter. The delta column is what that single choice is worth across the whole text, measured on the Ḥafṣ/Uthmānī corpus.
| Rule | The question | Worth |
|---|---|---|
| Shadda ّ | Is a doubled letter one letter or two? Classical sources record both positions explicitly. | +23,016 |
| Alif wasla ٱ | Written but often not pronounced. A letter, or not? | −13,819 |
| Superscript alef ٰ | Pronounced but written as a mark. The simple script writes it as a full alif. | +9,838 |
| Tanwīn ً ٌ ٍ | Represents a pronounced nūn. Counted as a letter, or as vowelling? | +8,893 |
| Lām-alif لا | One ligature or two letters — a live question in classical letter-counting. | varies |
| Disjoined letters | Is الٓمٓ three letters, or three named letters spelled out? | varies |
U+0621 to U+064A — silently drops ٱ alif wasla at U+0671, which sits above the range. That is 13,819 letters vanishing without any error. Classify by Unicode category, never by a guessed range.
Smaller in effect than the letter rules, but the same kind of choice.
| Rule | The question |
|---|---|
| Attached particles | فِي السَّمَاءِ — classical sources record it counted as one word by some, two by others. |
| The definite article | Is ال part of the word or a separate unit? |
| Prefixed conjunctions | وَ and فَ are written joined but are grammatically separate. |
| Orthographic joins | The two scripts differ on 367 words across the text — 77,881 against 78,248 whitespace tokens. |
"Classical" does not mean one agreed figure. It means several named men in the first two centuries who counted, and got different answers — and a tradition that preserved the disagreement rather than hiding it.
| Counter | Letters | Words |
|---|---|---|
| Sallām Abū Muḥammad al-Ḥimmānī | 340,740 | — |
| al-Ḥajjāj's commission — Ḥasan al-Baṣrī, Abū al-ʿĀliya, Naṣr b. ʿĀṣim, ʿĀṣim al-Jaḥdarī, Mālik b. Dīnār. Four months, counting with barley grains. | 323,015 | 77,439 |
| Abū Muʿādh al-Naḥwī | 321,200 | — |
| Mujāhid (d. 104 AH) | 321,000 | — |
| ʿAṭāʾ b. Yasār (d. 103 AH) | — | 77,437 |
| Ibn Masʿūd (d. 32 AH) | 300,670 | — |
These are transmitted in the standard works of ʿulūm al-Qurʾān, above all al-Suyūṭī's al-Itqān (d. 911 AH), which collects them without adjudicating. The spread on letters is 40,070 — 13%.
Multiplying the layers rather than guessing.
| Layer | Choices | Running total |
|---|---|---|
| Verse-counting system | 7 | 7 |
| Orthography | 2 | 14 |
| Letter rules (4 binary, minimum) | 16 | 224 |
| Word rules (2 binary, minimum) | 4 | 896 |
| × canonical transmissions | 20 | 17,920 |
One sentence, and everything downstream depends on it.
There is one strong consequence. Any pattern that appears under one method and vanishes under another is a property of that method, not of the text. So the honest test is to run every hypothesis under at least two verse systems and two orthographies, and report all four results — including the ones that fail.
A pattern that survives all four is worth attention. One that needs a particular twentieth-century typesetting to work is telling you about the typesetting.