On Its Own Terms

Reference analysis · 22 min read

The Historical Formation of the Hadith Corpus

On this page

This document sets out the principal historical-critical arguments concerning the origins of the hadith corpus and, more briefly, the textual history of the Quran. It is written from the perspective of academic historiography — asking what can be established about these texts using the methods applied to any other body of ancient literature, independent of the theological claims made about them.

The traditional Islamic sciences of hadith criticism are internally rigorous, but they operate on different premises and ask different questions. Where the two approaches diverge, the disagreement is usually methodological rather than a matter of one side possessing evidence the other lacks.

The arguments below vary considerably in how well established they are. Where a claim represents a leading position that remains contested, this is indicated.

A further disclosure is warranted. This document presents the revisionist case with its principal counter-arguments acknowledged and then judged insufficient to displace it. That judgement — that the critical scholarship narrows the gap between the corpus and its claimed origin without closing it — is defensible on the evidence but is a judgement rather than a neutral finding. A scholar persuaded by Motzki would arrange the same material to different effect, and would not be doing so in bad faith. Readers should treat what follows as a well-supported position within a live dispute, not as the settled consensus of the field.

The Transmission Problem

Hadith are transmitted with an isnad — a chain of named narrators purporting to run from the compiler back through successive generations to a Companion of Muhammad, and thence to the Prophet himself. The traditional sciences (mustalah al-hadith) developed a sophisticated apparatus for evaluating these chains: assessing each narrator’s reliability and character (jarh wa ta’dil), verifying that the chain is continuous, and checking whether the transmitters could plausibly have met.

This apparatus evaluates narrators individually. It does not analyse the structure of the transmission network as a whole. That structural question is where modern critical scholarship has concentrated.

When all recorded isnads for a given hadith are mapped as a network, a recurring topology emerges. Rather than radiating outward from the Prophet through many independent early strands, the chains frequently pass through a single transmitter living generations — often around a century — after Muhammad. Below this figure — the common link — the chains are thin, often a single strand. Above him, they branch widely in the pattern one would expect from genuine circulation.

Joseph Schacht identified this pattern in The Origins of Muhammadan Jurisprudence (1950); G.H.A. Juynboll developed the analysis substantially in subsequent decades. Their inference is that the common link marks the point at which the tradition entered circulation. The chain reaching back beyond him represents a later construction supplying prophetic authority to material that originated with, or shortly before, the common link himself.

Traditional hadith criticism cannot detect this pattern, because the pattern is a property of the network rather than of any individual narrator. Each link in a back-projected chain may consist of individually reputable figures.

An illustrative case

The hadith innama al-a’mal bi’l-niyyat — “actions are judged by their intentions” — opens Sahih al-Bukhari and is among the most widely cited in the corpus.

The tradition itself classifies it as gharib: transmitted through a single chain. That chain runs Umar ibn al-Khattab → Alqama ibn Waqqas → Muhammad ibn Ibrahim al-Taymi → Yahya ibn Sa’id al-Ansari (d. c. 143 AH). At Yahya, transmission expands abruptly to a large number of narrators.1 Sahih al-Bukhari, Kitab Bad’ al-Wahy, narrated Umar bin al-Khattab; tr. Muhsin Khan (full text). The Arabic opens innama al-a’mal bi’l-niyyat, “actions are only by intentions”; “The reward of” is the translator’s gloss. The single early chain (Umar → Alqama ibn Waqqas → Muhammad ibn Ibrahim al-Taymi → Yahya ibn Sa’id al-Ansari) is the one recorded in Bukhari’s own isnad for this hadith: “I heard Allah’s Messenger saying, “The reward of deeds depends upon the intentions and every person will get the reward according to what he has intended. So whoever emigrated for worldly benefits or for a woman to marry, his emigration was for what he emigrated for.""

Classical scholars observed the singularity of the early chain and discussed it under the category of gharib. The structural reading takes the observation further: a saying of this significance transmitted by exactly one person per generation for approximately 130 years, then suddenly by dozens, is more economically explained by origination at the point of expansion than by an improbably narrow bottleneck preceding it.

This inference is probabilistic rather than demonstrative. A saying may genuinely have originated with the Prophet, been heard by few, transmitted quietly through a narrow line, and only later acquired the currency that produced wide transmission. Nothing in the isnad structure alone excludes this. The revisionist case instead holds that such narrow survival is improbable for material of this importance and recurs too systematically to be repeated coincidence. The shape of the evidence supports that argument, but does not prove it. This particular hadith is also the standard illustration against which defenders of the tradition have written most directly.

The limits of this argument

The common link thesis is the leading revisionist account, but it is not uncontested.

Harald Motzki’s isnad-cum-matn method examines chains and texts together, asking whether textual variants across branches correlate with the points at which the chains diverge. Where they do, this suggests genuinely independent transmission rather than a single fabricator generating multiple chains. Applying this to material in Abd al-Razzaq’s Musannaf and elsewhere, Motzki has argued that certain traditions can be dated earlier than their apparent common links. Gregor Schoeler has developed related arguments concerning the interplay of oral and written transmission in the early period, suggesting the absence of early written collections is less anomalous than Schacht supposed. From within the tradition, M.M. Azami mounted a sustained critique of Schacht’s chronology and use of sources.

The strongest form of the objection is Motzki’s critique of the common-link inference itself. On Motzki’s reading, the common link is not the point of origination but the first systematic collector: an early specialist who gathered material and disseminated it professionally, whose activity is why recorded transmission begins with him. The single strands below the common link are then artifacts of how transmission was recorded — a collector named his principal source, not every path the material had traveled — and the network’s shape cannot by itself distinguish a fabricator from a collector. The claim is underdetermination, not authenticity: Motzki did not argue that the strands below the common link are shown to be genuine, only that the topology does not show they are not. This position and its own limits are set out under “The Field Since 2010” below.

Two observations about this counter-scholarship:

First, its successes generally push dating back to the Companion or senior Successor generation rather than to Muhammad himself. The gap narrows; it does not close.

Second, the common-link pattern itself has not been displaced as the general structural description of the corpus. What the collector reading contests is what the pattern licenses one to infer; between origination and first systematic collection the question is genuinely open, and Juynboll’s reading remains the more economical for many transmission bundles.

The calibrated position is that common links are a real and widespread feature whose interpretation remains debated. The reviewed evidence does not justify a universal negative about every hadith under one set of “ordinary historical standards.” It does show that traditional grading and network topology alone do not establish prophetic origin; each attribution requires report-specific historical argument.

The Direction of Attribution

A separate line of argument concerns how attributions changed over time.

Schacht observed that comparing legal texts of different dates reveals the same substantive ruling attributed to progressively more authoritative figures. A ruling appearing in early Kufan material as the opinion of the local jurist Ibrahim al-Nakha’i (d. c. 96 AH) appears in later material as the opinion of Ibn Mas’ud, the Companion associated with Kufa, and later still as a prophetic hadith with a complete chain.

Malik’s Muwatta exhibits a comparable pattern internally. Malik frequently presents rulings simply as the settled practice of Medina, without prophetic attribution. Some of these same rulings appear in the later canonical collections as prophetic statements furnished with full isnads.

A further category of case involves legal maxims whose final form places them in maximally authoritative narrative settings. The inheritance maxim la wasiyya li-warith (“there is no bequest to an heir”) reaches the canonical collections as a pronouncement delivered during the Farewell Sermon — the most authoritative setting available in the biographical tradition — while the underlying rule is also discussed in the juristic literature as a doctrine argued on its own terms and connected to the Quranic inheritance shares. Whether this represents a documentable escalation of the kind traced above, or a genuinely early prophetic ruling that jurists also defended independently, depends on source chronology that should be verified case by case rather than assumed from the shape of the pattern.

This caution applies generally. The escalation argument is strongest where the relative dating of the compared texts is independently secure — as in the Kufan material above — and weakest where a source’s date is itself inferred from its contents.

Within that constraint, the observed movement in the documentable cases is consistently toward higher authority, and the reverse movement is not a notable feature of the material. This is what one would expect if the operative mechanism was competitive legitimation rather than progressive recovery of better sources. It should be read as a strong tendency identified across Schacht’s survey and its successors, not as an established universal.

This argument carries a methodological dependency worth stating plainly: it rests entirely on the relative dating of the texts being compared. Where that chronology is secure, the inference is strong. Where the dating of a source is itself inferred from its contents, there is a risk of circularity, and critics of Schacht have pressed this point.

The garrison towns of the expanding empire — Kufa, Basra, Fustat, Damascus, alongside Medina — faced immediate administrative demands: taxation, land settlement, distribution of conquest revenue, marriage across newly mixed populations, commerce with non-Muslim subjects, criminal jurisdiction. Local legal practice (amal) developed to meet these needs, initially resting on the authority of local scholars and governors.

As regional schools consolidated and came into competition, local authority became insufficient to settle disputes between them. The available escalation ran from local jurist to Companion to Prophet. Once any school anchored its positions in prophetic hadith, others faced pressure to match.

On this account the fabrication was not primarily conspiratorial but institutional: a competitive dynamic in which each escalation was locally rational and the cumulative outcome was a prophetic corpus constructed to adjudicate second- and third-century legal questions.

Whether the participants understood themselves to be fabricating is a separate matter from what was produced. The Quran repeatedly condemns the invention of falsehood against God, and the corpus itself preserves the warning that whoever deliberately lies about the Prophet has secured his place in the Fire.2 Sahih al-Bukhari, Kitab al-‘Ilm, narrated Abu Huraira; tr. Muhsin Khan (full text). The parenthetical “(intentionally)” renders the Arabic muta’ammidan, “deliberately,” which is present in the text: “And whoever tells a lie against me (intentionally), then (surely) let him occupy his seat in Hell-fire.” The tradition additionally developed an entire science of forged traditions (mawdu’at), demonstrating that deliberate fabrication was recognised as widespread and was understood as a grave offence. Attributing specific words to Muhammad on no evidence that he spoke them falls within the prohibition the participants themselves acknowledged.

The Absence of Contemporary Documentation

No hadith survives in a document contemporary with Muhammad, and none is independently corroborated by non-Islamic sources of the period.

The customary explanation invokes an oral culture in which written records were not the norm. This explanation sits uneasily with the tradition’s own account of the period. That account describes Quranic revelation being recorded during the Prophet’s lifetime on parchment, bone, and palm stalks;3 Sahih al-Bukhari, Kitab al-Tafsir, narrated Zaid bin Thabit; tr. Muhsin Khan (full text). Abu Bakr commissioning the first collection: “You are a wise young man and we do not suspect you (of telling lies or of forgetfulness): and you used to write the Divine Inspiration for Allah’s Messenger. Therefore, look for the Qur’an and collect it (in one manuscript). […] So I started locating Qur’anic material and collecting it from parchments, scapula, leaf-stalks of date palms and from the memories of men (who knew it by heart).” presents the Constitution of Medina as a written instrument; describes the Treaty of Hudaybiyyah as drafted in writing, with a dispute over its wording;4 Sahih al-Bukhari, Kitab ash-Shurut, narrated al-Miswar bin Makhrama and Marwan; tr. Muhsin Khan (full text). “So, the Prophet called the clerk and said to him, “Write: By the Name of Allah, the most Beneficent, the most Merciful.” Suhail said, “As for ‘Beneficent,’ by Allah, I do not know what it means. So write: By Your Name O Allah, as you used to write previously.” […] Then he dictated, “This is the peace treaty which Muhammad, Allah’s Messenger has concluded.” Suhail said, “By Allah, if we knew that you are Allah’s Messenger we would not prevent you from visiting the Kaba, and would not fight with you. So, write: "Muhammad bin Abdullah."" and reports written correspondence with foreign rulers. Mecca is portrayed as a commercial centre where written contracts were in use.

Writing was therefore available and was applied to matters considered important. The tradition nonetheless preserves no contemporary record of the sayings that would later be elevated to the second source of law.

This leaves three readings, not two. Either the sayings were not recorded because they were not yet regarded as authoritative in the way later doctrine holds them to be, which weakens their subsequent theological status; or they did not yet exist in the form the tradition presents, which supports reconstruction; or they were authoritative and writing was deliberately withheld to protect the Quran’s exclusivity. The third is the tradition’s answer: Sahih Muslim, Kitab al-Zuhd, carries the Prophet’s instruction not to write down anything from him other than the Quran, to erase what had been written, and to narrate from him orally.5 Sahih Muslim, Kitab al-Zuhd wa’l-Raqa’iq, narrated Abu Sa’id al-Khudri; tr. Abdul Hamid Siddiqui (full text). An aside by the transmitter Hammam, recalling that the word “deliberately” was also said, is trimmed: “Do not take down anything from me, and he who took down anything from me except the Qur’an, he should efface that and narrate from me, for there is no harm in it and he who attributed any falsehood to me […] he should in fact find his abode in the Hell-Fire.” Permission reports also exist — Abdullah ibn Amr’s writing and the sermon written for Abu Shah — alongside Umar’s fear of producing a book beside the Book; harmonizations distinguish periods or persons. The answer is real, but repeated permissions leave it only a partial explanation of the documentary gap.

The third reading rests on reports that are themselves hadith: the corpus explains its origins through the very channel in question, so the explanation cannot stand on firmer ground than the material it explains. It also coexists with contrary permission reports, which the harmonisations accommodate rather than dissolve.

Schoeler’s work complicates this somewhat by arguing for a hybrid oral-written transmission in which written notes served as memory aids within a fundamentally oral system, leaving a documentary trace that would not be expected to survive. This weakens the argument from silence without dissolving it: notes-as-aide-mémoire still do not produce the contemporaneous authoritative record that the theological claims about hadith would lead one to expect.

The general point stands independently of how one weighs these readings. The entire edifice of isnad criticism exists because direct documentary evidence is absent. Where contemporary records survive, chain-of-narrator authentication is unnecessary.

The Quranic Text

The Quran occupies a different evidentiary position from the hadith corpus and should not be assimilated to it.

Manuscript evidence for the Quran is substantially better. Early manuscripts — the Birmingham folios, the Sana’a palimpsest, the Codex Parisino-petropolitanus and others — support the circulation of Quranic material in written form within the first century of the Islamic era, close to the traditional chronology. The Quran is not a second-century composition in the way that much of the hadith corpus appears to be.

One distinction should be preserved here, since it is frequently collapsed in popular treatments. Radiocarbon analysis dates the parchment, not the ink. The Birmingham folios yielded a range of 568–645 CE at 95% confidence, whose early bound sits awkwardly against the traditional chronology of revelation and has generated corresponding debate; palaeographic dating of the script generally places the writing somewhat later than the outer bound of the radiocarbon range. The material evidence therefore establishes early parchment bearing Quranic text, which supports an early written tradition but does not by itself demonstrate a fixed text at that date.

The stronger traditional claim — perfect preservation — is harder to sustain.

The Uthmanic recension, as the tradition itself describes it, involved establishing a standard text and destroying divergent copies. That is an editorial and political act. Whatever its fidelity to earlier material, it is a process of standardisation rather than passive transmission, and it necessarily removed the evidence by which its own fidelity might be assessed.

The Sana’a palimpsest, examined by Behnam Sadeghi and Mohsen Goudarzi, preserves a lower text differing from the standard Uthmanic text in ways that go beyond orthography — including variations in wording and in the ordering of material. It appears to represent a non-Uthmanic textual tradition that was subsequently effaced.

The canonical qira’at traditions preserve genuine differences in reading. The tradition accounts for these as divinely sanctioned modes of recitation. An alternative reading treats them as residue of a text that had not been entirely fixed at the point of standardisation, subsequently accommodated within an authorising framework.

The resulting picture is of an early text — genuinely early — that passed through an editorial process backed by political authority, producing a standardised version at the cost of alternatives now largely irrecoverable. This is a substantially different claim from perfect preservation, though it is also a substantially stronger position than that occupied by the hadith literature.

The Field Since 2010

The positions summarized above hardened into place by the 1990s. The work since roughly 2010 has not settled the dispute, but it has changed its shape — and it cuts both ways. Neither direction of what follows is a courtesy concession.

The common link reread. Motzki’s mature position is not a case-by-case redating but a critique of the common-link inference itself (“Dating Muslim Traditions: A Survey,” Arabica 52 (2005); Analysing Muslim Traditions, with Nicolet Boekhoff-van der Voort and Sean W. Anthony, 2010). On his reading, the common link marks the first systematic collector — an early specialist who gathered material and disseminated it professionally — and the single strands below him are artifacts of recorded transmission: a collector named his principal source, not every path the material had traveled. The network’s shape cannot by itself distinguish origination from first systematic collection. The claim is underdetermination, not authenticity — Motzki does not hold that the strands below the common link are shown genuine, only that the topology does not show they are not. Against it stands the economy of Juynboll’s reading for many transmission bundles; the collector reading is itself an inference to a possibility, and the question is genuinely open.

The ‘Urwa corpus. Andreas Görke and Gregor Schoeler applied the isnad-cum-matn method to the corpus attributed to ‘Urwa ibn al-Zubayr (d. c. 94 AH), Aisha’s nephew, transmitted through al-Zuhri and through ‘Urwa’s son Hisham (Die ältesten Berichte über das Leben Muhammads, 2008). Where independent lines converge, they reconstruct a core of accounts — the beginning of revelation, the hijra, the ifk affair, Hudaybiyya — taught in Medina in the late first Islamic century. The claim concerns the age of the accounts, not their historicity. Stephen Shoemaker’s response (“In Search of ‘Urwa’s Sira,” Der Islam 85 (2011)) found the reach back to al-Zuhri credible and the further reach to ‘Urwa generally unconvincing — single strands, judgment calls — and the authors replied in a joint debate piece (Der Islam 89 (2012)). “The sira has been dated to the first century” would overstate what survives the exchange; the dating is contested at precisely the ‘Urwa step.

A check that does not trust the chains. Behnam Sadeghi’s traveling-tradition test (“The Traveling Tradition Test,” Der Islam 85 (2008)) dates material without trusting individual isnads: in three case studies, textual variants cluster by the cities named in their chains — evidence that isnads carry genuine geographic and historical information, whatever their theological freight. It is a statistical check on the instrument, independent of confidence in any single chain.

The method in stricter hands. Pavel Pavlovitch’s study of the kalala traditions (The Formation of the Islamic Understanding of Kalala, 2016) ran the isnad-cum-matn method across twenty-nine tradition clusters and dated most archetypes to the second century AH — the method, rigorously applied, frequently yields later datings than Motzki’s own applications did. The kalala material is unusually contested, and the result should not be generalized in either direction.

The hinge case. Joshua Little’s Oxford doctoral thesis (2022) applied the isnad-cum-matn method to every version of the hadith of Aisha’s marital age and concluded that the archetype was formulated by Hisham ibn ‘Urwa (d. c. 146 AH) and disseminated after his move to Iraq, in an Abbasid-era polemical context. Its status here is an examined doctoral thesis whose archetype argument contains contestable judgment calls. Its methodological significance is limited but real: a method used to argue for early dating produced a late, skeptical result in this case, so the method does not mechanically guarantee either conclusion.

The external floor. Sean Anthony (Muhammad and the Empires of Faith, 2020) assembles the earliest non-Islamic evidence — the Doctrina Iacobi and the chronicle notice of Thomas the Presbyter, both within roughly a decade of Muhammad’s death, with coins, inscriptions, and papyri behind them — establishing Muhammad as a documented prophet-figure far earlier than the Islamic literary sources were compiled, and argues that critically handled sira-maghazi material preserves early layers. The floor is firm and its limit must be kept: the external sources corroborate the man — his existence, his period, his prophetic claim — not the content of the hadith corpus. The former cannot be spent to underwrite the latter.

The Quran side. Behnam Sadeghi and Uwe Bergmann’s study of the Sana’a palimpsest’s lower text (“The Codex of a Companion of the Prophet and the Qur’an of the Prophet,” Arabica 57 (2010)) radiocarbon-dated the parchment to 614–656 CE at 68% probability — 578–669 at 95%, with a 75.1% probability of predating 646; dates for the parchment, not the ink — and argued a stemma on which the C-1, Uthmanic, and Ibn Mas’ud text types descend from a common prototype, which the authors identify with the Prophet’s own recitation. Their further conclusions are argued inferences from that stemma, not laboratory results: on their analysis, the Uthmanic tradition reproduces the prototype more faithfully than C-1 does, and the ordering of verses and sentences was already fixed in the prototype. Asma Hilali’s rival reading of the palimpsest as a scribal exercise (The Sanaa Palimpsest, 2017) was strongly criticized by Nicolai Sinai’s review in the Journal of the American Oriental Society and Éléonore Cellard’s material study (“The San’a’ Palimpsest: Materializing the Codices,” Journal of Near Eastern Studies 80 (2021)), which finds a complete, professionally produced codex. Those responses support saying Hilali’s thesis has not become the dominant account; they do not establish that it attracted no scholarly following.

Where a widely used survey stands. Jonathan Brown’s Hadith: Muhammad’s Legacy (2nd ed., 2017) presents the classical Sunni hadith sciences with technical detail alongside an account of the Western critical debate. Its stance is sympathetic to the traditional apparatus while conceding critical findings at points; that methodological calibration, rather than the author’s personal identity, is what matters here.

The net effect for the judgment this document carries is case-specific. Isnad-cum-matn studies often reach the Companion or Successor generation, while some scholars argue further back in particular cases and skeptical results can emerge from the same method. The gap narrows unevenly; no corpus-wide rule makes prophetic attribution automatic or impossible.

Consequences

The scope of the problem depends on how much of Islamic practice rests on hadith rather than on the Quran.

The Quranic text does not specify the form or number of the daily prayers, the rates and thresholds of zakat, most of the mechanics of inheritance beyond the fixed shares, the procedures of marriage and divorce in detail, the prescribed punishments for several offences including apostasy, or the greater part of ritual and dietary law as practised. This material derives from hadith and from the juristic tradition built upon it.

If the revisionist historical-critical account is broadly correct, the consequence is not that Islam lacks a foundation. The Quran remains, and its early dating is well supported. The consequence is that much of the practised religion — its law, ritual forms, and daily obligations — rests on attributions whose prophetic origin remains historically disputed and whose recorded forms often become visible only in the second and third Islamic centuries. This does not establish that the whole practised religion is historically untraceable to Muhammad.

What remains is an early text of demonstrable antiquity that underwent a human editorial process, surrounded by a legal and ritual apparatus of substantially later construction. That is a coherent object of historical study. It is not the object described by the tradition’s account of itself.

Principal Sources

Revisionist:

Critical of the revisionist position:

The field since 2010 (see the section of that name):

Quranic textual history: