A compilation of expert Twitter threads and tweets on the Qiraʼat readings and textual transmission of the Quran

By Avnar Sidiche

This page is intended as a useful overview of the most interesting points shared by academics during the recent perfect preservation controversy of 2020.

Concepts of preservation

“Preservation” is not well defined, but Islamic scholarship saw no contradiction from the existence of variant readings.

Word-for-word preservation is closer to a defensible doctrine (contestable due to variant readings, though traditionally these are accommodated).

The most dogmatic preservation claim “requires an extraordinary amount of special pleading”.

The case that all the canonical variants were ahruf uttered by the Prophet is “close to the opposite of undeniable”.

An Uthmanic rasm standard (skeleton text without diacritics) is likely

The stemmatic tree relationship (Michael Cook) of documented regional rasm variants, which accords well with manuscript evidence (Yasin Dutton – click and scroll up; see also Hythem Sidky’s analysis) is strong evidence supporting the traditional account of an Uthmanic archetype sent out to certain cities.

Further supported by radio-carbon dating.

[account deleted]

A thread explaining the methodology with examples of manuscripts dated to the 7th century.

[account deleted]

The Uthmanic rasm is well preserved (not perfectly).

It seems that the changes were minor (spelling). The Cairo edition is an attempt to reconstruct it.

The rasm is not the whole picture in terms of preservation. Many possible readings can be imposed upon it and we find compatible non-canonical readings in manuscripts too.

The canonical readings also sometimes depart from the standard rasm (see below).

Some canonical readings are not found in early manuscripts

Some canonical readings (including the Ḥafṣ ʿan ʿĀṣim transmission) are not found in the earliest manuscripts.


and a more detailed thread on the lack of the reading of Hafs in early manuscripts. Variants unique to Hafs (among canonical readers)  rarely appear as a primary reading in early manuscripts, even as part of non-canonical readings.

It’s surprising that al-Ṭabarī had never heard any variants unique to Ḥafṣ (click and scroll down for example demonstrating this) if canonical variants were mutawātir (mass-transmitted like the rest of the Qurʼān), as claimed by later scholars.

From an academic perspective, it is unknown how far back the readings really go beyond the eponymous readers.

Were the canonical readings always regarded as perfectly transmitted?

Ṭabarī (and Zamakhsharī later) is the first scholar known to criticise variants found in the canonical readings before they came to be regarded as unquestionable. This thread goes through highlights of the first 47 pages of Shady Nasser’s book on the transmission of the readings. My note: see also pages 58–61 on criticisms by others and efforts to justify variant readings.

[account deleted]

Even Ibn Mujāhid (who canonised the seven readings) criticised some canonical variants.

Other examples can be found be searching for the word “wrong” in the below open access book appendix by Prof. Shady Nasser. It lists all the variants among the seven readings recorded by Ibn Mujāhid, who sometimes regarded a canonical variant as wrong (not just a transmission thereof).


Early grammarians such as al-Farrā criticised canonical variants even before al-Ṭabarī, though theirs was a quite different discipline.

Are canonical reading variants pre-Uthmanic?

Canonical readings have a close correspondence with the Uthmanic rasm variants present in their regions. Evidence indicates that the canonical readings were adapted to fit these regional variant codices rather than the other way around (see whole thread; Marijn’s Hišām’s ʾIbrāhām paper gives further evidence).

These regional textual variants seem to have been scribal errors.

Hythem Sidky (see his 2020 paper listed on the Resources page) gives many further reasons why these regional variants are scribal errors.


Far more variants seem to have arisen from ambiguities where multiple possible words fit the same rasm (click and scroll up for best example).

Though lack of variants in most cases of ambiguity demonstrates that there was consensus on how to read most of the Qurʼān.

And reading variants are probably not just down to the ambiguous rasm and its regional variants.

Transmitters of the same eponymous reader also differed from one another.

A longer thread on the topic of transmission.

The canonical readings were not in the Qurashi dialect

A longer thread on these dialects and the paradox that in contrast, the Qurʼān does look to have been originally in the Qurashi dialect.

See also Marijn’s papers showing that the spoken Qurʼān originally lacked use of hamza except with word-final ā (Hamzah in the Quranic Consonantal Text) and had a reduced grammatical case system vs. the readings we have today (Case in the Qurˀānic Consonantal Text).

Non-canonical readings

The lower text of the Ṣanʽā’ 1 palimpsest, our only known example of a non-Uthmanic text type, contains readings which in the literature were attributed to various Companions.

The reading of ibn Masʿūd, popular in Kufa at one time, is a well known example and did not comply with the Uthmanic rasm, and there is good corroboration that it really existed.

Non-canonical readings on the Uthmanic text type are by far the most common in early manuscripts (see also above).

Do the pattern of variants fit with a Prophetic origin?

In this last section I am just asking my own questions as a lay person and making some inferences.

Hundreds of variants affect meaning in small ways and are not just down to dialects/spelling. If all canonical variants were uttered by the Prophet and did not arise later, we can ask whether he just recited ahruf variants intended to fit a single rasm (when eventually written) or in many?

If he recited ahruf intended to fit a single rasm when written down, why are there some exceptions? Even canonical reading variants don’t always comply with the Uthmanic rasm (nor its regional textual variants). See this thread:

My note: The example given in the thread from 19:19 is part of a quote of what someone said. It seems likely that such a conflicting story between the variants would arise after Muhammad’s career. The 19:19 variants sound similar, so could also have arisen during oral transmission. There were also a lot of Companion readings and other non-canonical readings that do not fit the rasm (see above).

If instead Muhammad recited in ahruf variants that were unrestrained by a particular written rasm, of which the canonical variants today represent a small subset, then tens of thousands of variants that would fit other rasms (both meaningful and dialect/spelling differences) would later have been discarded for the sake of a standardised rasm (not in order to limit to one dialect). It seems unlikely standardisation would have succeeded if such a vast number of variants did not fit the Uthmanic rasm.

This leaves the possibility that fits all the evidence (including the limited early attestation and early criticisms mentioned above) — that while there was some variation in pre-Uthmanic times, many of the reading variants post-date the Uthmanic standardisation. It looks very much like many arose from ambiguity in the rasm and from its regional textual variants as well as phonic similarities during oral transmission.