Did Muhammad Exist?

Summary

A small but loud wing of Christian apologetics argues that Muhammad never existed — that the prophet of Islam is a legend assembled in the eighth and ninth centuries out of an Arab empire that had no founder.  This article tests that claim by the same rules the Berean Archive uses on Jesus mythicism.  The claim fails, and it fails for the same structural reason.

• The sceptical case has one strong argument, and it is a real one. Muhammad's name does not appear on a single dated Arabic coin, inscription, or state document until the 680s — roughly fifty-four years after his traditional death in 632. Robert Hoyland, who catalogued the evidence, says so himself: "before AH 72 the archaeological record is strangely silent about Islam, and this despite the fact that we do have a fair amount of material from this time."¹ That is the whole case, and this article states it at full strength before answering it.
• But the silence is only in Arabic, and only in official media. Outsiders were writing about him from 634 onward. A Greek tract composed in July 634 reports that "a prophet has appeared, coming with the Saracens," preaching the coming Messiah and claiming "the keys of paradise."¹ That is about two years after the traditional death date. The prophet is not named.
• The name appears within five years. A Syriac note scribbled around 637 on the flyleaf of a sixth-century Gospel book names Muhammad twice, in a scrap that also dates the battle of Gabitha to 20 August 636.² Sean Anthony calls it "the earliest extant document to mention Muḥammad tout court by name."³
• Eight years to a notice whose date passes three independent checks. Thomas the Presbyter, writing about 640, records "a battle between the Romans and the Arabs of Muḥammad" on Friday 4 February 634, twelve miles east of Gaza.¹ The Seleucid year, the Roman indiction, and the Julian weekday all agree. Fabricating a battle is easy. Fabricating one whose three independent calendars line up is not.
• The evidence keeps coming, in four languages, from writers with no motive to help. An Armenian chronicle of 655 gives Muhammad's preaching at length and names its source as prisoners of war.⁴ A Syriac chronicler in Iraq in the 660s calls him the conquerors' leader.¹ John bar Penkaye in 687 calls him their "guide" and "teacher" whose laws they enforce on pain of death.² None of these men liked the Arabs. None had seen a Quran.
• The documentary trail runs in parallel. An Egyptian tax receipt of April 643 is dated by a calendar counted from an emigration nobody in the seventh century ever explains.³ A graffito of AH 24 near al-Hijr dates itself "at the time ʿUmar died" and matches the caliph's traditional death date to within about a day.³
• The mythicist case needs every one of these explained away separately. The Greek tract is a rumour. The flyleaf is a title, not a name. Thomas was interpolated. Sebeos is legend. The calendar means something else. Any one of those moves is arguable. Needing all of them at once is the mark of a theory protecting itself.
• This is a case where the Christian apologetics community is divided, and one wing of it is wrong. Robert Spencer's Did Muhammad Exist? and Jay Smith's presentations argue the negative. David Wood argues the affirmative, against both of them.⁵ The evidence is on Wood's side, and saying otherwise costs a Christian reader nothing but costs the argument its credibility.
• The honest weaknesses of this article's case. The earliest witnesses are outsiders reporting hearsay, and Nicolai Sinai concedes they are probably repeating what Muslims told them.⁶ The earliest Syriac notices give Muhammad no religious role at all — Anthony says they "exhibit not a shadow of cognizance that the conquests were inspired either by a prophet or, for that matter, any religious message at all."³ The traditional death date of 632 is not secure; several early sources put Muhammad alive at the invasion of Palestine in 634. No seventh-century source outside Arabia mentions Mecca, which Patricia Crone treated as an unsolved problem to her death.⁷ And the detailed biography — the cave, the Night Journey, the battles, the wives — rests on books written 150 to 250 years later, which is a separate and much weaker case than the case for his existence.

Verdict:  a man named Muhammad led an Arabian monotheist movement in the early seventh century, and his followers conquered the Near East in his name.  That is about as well established as anything in seventh-century history, and it is better attested from outside than Jesus is.  The mythicist case rests on a genuine anomaly — the fifty-four-year Arabic silence — and then makes that anomaly do work it cannot do, because the same fifty-four years are full of non-Arabic writing about the same man.  The anomaly has a better explanation than non-existence:  the public, official proclamation of Muhammad was a political act of the second Islamic civil war, first used by rebels against the Umayyads and then adopted by them.  What the evidence does not establish is the traditional biography.  Christians who want to argue against Islam have a strong case about the sources for Muhammad's life and a weak one about whether he lived.  Choosing the weak one is a mistake.

The claim being tested

"Did Muhammad exist?" hides at least five separate questions.  Popular argument on both sides runs them together, and that is where most of the confusion comes from.  Separate them:

1. Did a man named Muhammad live in western Arabia in the early seventh century? This is the bare historicity question.
2. Did he lead a monotheist movement whose members later conquered the Near East? This adds a public role.
3. Is the traditional biography — the sira, the life of Muhammad — reliable history? The sira is the narrative biography, and it is a different body of writing from the hadith, the collected reports of what Muhammad said and did. This asks whether the cave on Mount Hira, the Night Journey, the battle of Badr, and the marriage to Aisha are events or literature.
4. Was he based in Mecca, and is the Mecca of the tradition the Mecca on the map?
5. Is the Quran the collection of his own preaching?

A Muslim answers yes to all five, and treats them as one claim.  The sceptics this article examines answer no to all five, and also treat them as one claim.  Both are wrong to bundle them.  The evidence for (1) and (2) is early, external, and in four languages.  The evidence for (3) is late, internal, and weak, and the field says so out loud.  Question (4) is genuinely open.  Question (5) is answered elsewhere on this site, in the article on the Quran's origins and the article on its preservation.

This article decides (1) and (2), reports honestly on (3) and (4), and leaves (5) alone.

The precise claim being tested is therefore the one Robert Spencer states at the end of his book:  the historical record "suggest[s] that the Muhammad of Islamic tradition did not exist, or if he did, he was substantially different from how that tradition portrays him."⁸  Note the two halves.  The second half — "substantially different from how that tradition portrays him" — is close to the mainstream position and this article largely agrees with it.  The first half is the claim under test, and it fails.

The same test used on Jesus

The Berean Archive has already run this test once, on the claim that Jesus of Nazareth never existed.  Running it a second time with different rules would be worthless.  So the rules are the same rules, stated here so a reader can check that they are.

• State the sceptical case from its own sources, at full strength, before answering it. The Jesus article quotes Richard Carrier's own numbers and calls his theory "not absurd." This one quotes Spencer, Jay Smith, Yehuda Nevo and Judith Koren, and Karl-Heinz Ohlig from their own books.
• Weigh the earliest independent evidence first, and say how many years separate it from the events. For Jesus that is Paul, writing about twenty years after the crucifixion, who says he met Jesus' brother. For Muhammad it is a Greek tract of 634 and a Syriac flyleaf of about 637.
• Do not let an argument from silence carry a conclusion it cannot carry — in either direction. The Jesus article argues that the silence of first-century writers is expected, because those writers mention none of the comparable figures either. The Muhammad case is different and the difference must be conceded: the seventh-century Arabic record names caliphs and governors freely, so the silence about Muhammad is not expected and needs a real explanation. This article gives one.
• Count how many separate rescues the sceptical theory needs. A theory that must explain away each piece of evidence by a different mechanism is doing what the Jesus article called "protecting itself from evidence."
• Retire our own side's bad arguments in public. The Jesus article ends by retiring several popular Christian proofs that fail checking. This one does the same, and the biggest item on the list is the claim under test.

One result of applying the same test twice is uncomfortable for a certain kind of apologetics, so it should be said plainly at the start:  the external, non-Christian evidence for Muhammad is earlier than the external, non-Christian evidence for Jesus.  Muhammad is named by an outsider within about eight years of his death.  Jesus is named by a non-Christian historian about sixty years after his.  The Christian case does not need the comparison to come out the other way, and asserting that it does is false.

The two records differ in shape, and that difference is the honest observation.  For Jesus, the earliest evidence is internal — a follower's letters from within twenty years — and the external evidence is late.  For Muhammad it is the reverse:  the external evidence is early and abundant, and the internal evidence is late and thin.  A skeptic should notice that these are two different failure modes and that neither one is fatal by itself.

The strongest case for the Muslim claim

A Muslim does not think this is a hard question, and it is worth stating why before any Western evidence is brought in.  The case, in Muslim terms, is cumulative.

A community that never lost the thread.  Islam did not have to rediscover its founder.  From the first generation there were men and women who said they had met him, and the reports of what they saw were passed on by name, from teacher to student, in an unbroken chain called an isnad — literally a "support," the list of who heard it from whom.  By the ninth century the scholars had built an entire science around auditing those chains:  who was reliable, who was careless, who could plausibly have met whom, whose memory failed in old age.  Al-Bukhari (d. 256/870) is reported to have selected his collection "from over 600,000 ḥadīths and spent ten years compiling it" — a figure that comes from a later admirer rather than from Bukhari, and that Jonathan Brown treats as part of the praise-literature that grew up around the book rather than as an audited statistic.⁹  The whole apparatus assumes a real man at the origin, and it was built by people much closer to him than any modern critic.

The Quran itself.  Whatever else is disputed, the Quran is an early text and it names Muhammad four times.  Sean Anthony, who is not a Muslim and is careful about what he grants, puts it strongly:  "The Qurʾan is the earliest and most important artifact of the life of Muḥammad … Moreover, the Qurʾan's documentation and the material evidence for its redaction and transmission are peerless in the Arabic literary corpus.  This assertion reflects, not the naïve sentiments of believers or pietistic scripturalists, but rather an emerging consensus based on over a century and a half of Western scholarship."³

A movement that behaved as though it had a founder.  Within a decade of the traditional death date the conquerors were dating documents by a calendar counted from an emigration, and calling themselves in Greek magaritai, a word coined from the Arabic muhajirun, "emigrants" — those who have made a hijra, a migration.³  A movement whose self-description is built on a departure presupposes someone who departed.

Modern source criticism has moved toward the tradition, not away from it.  This is the part most often missed.  The technique called isnad-cum-matn analysis compares many versions of the same report — matn means the text of the report, as against the chain — and works backward to the common source they all descend from.  Anthony summarises what it has achieved:  "the current consensus holds that, at the very least, we have a robust sense of what one of Ibn Isḥāq's teachers, the scholar Ibn Shihāb al-Zuhrī (d. a.h. 124/742 c.e.), transmitted about Muḥammad. … we even know what one of al-Zuhrī's teachers, ʿUrwah ibn al-Zubayr, likely said as well."³  Urwa died in 93/711 or 94/712.  That is inside eighty years of the events, not two hundred.

And the sceptical wing of the field is shrinking, on its own account.  Herbert Berg, himself a sceptic, wrote in 2003 that "a consensus is forming in support of the sanguine scholars, leaving (at least in the minds of the former) the burden of proof with the skeptical scholars."¹⁰  Twenty years later Andreas Görke and Gregor Schoeler put it more bluntly:  the debate "is still going on;  however, the partisans of the 'sceptics' have shrunk to a small group in the meantime."¹¹

That is a serious case and a believing Muslim should recognise it here without complaint.  It has one weakness:  every link in it except the Quran and the papyri is a source written down long after the fact.  The earliest surviving connected biography is Ibn Hisham's edition (d. 218/833) of Ibn Ishaq (d. 150/767), and the great hadith collections are later still.  Stephen Shoemaker's description is harsh but not eccentric — the traditional biographies are "widely recognized by modern historians as little more than pious imaginations about the beginnings of Islam that took shape in the collective memory among Muhammad's followers over at least a century after his death"² — and Fred Donner, who is far friendlier to the tradition, says the same thing more quietly:  "we have no original documents that might confirm unequivocally any of the traditional biography."¹²

So the Muslim case cannot be the whole answer, because it is mostly built out of the material whose reliability is exactly what the sceptic disputes.  Something independent is needed.  Fortunately there is a great deal of it.

The strongest case for the sceptics

The sceptical case is not silly and it is not built on nothing.  It has one central observation, which is true, and which the leading cataloguer of the evidence states in his own words:  "Yet before AH 72 the archaeological record is strangely silent about Islam, and this despite the fact that we do have a fair amount of material from this time."¹  Twenty years later Hoyland said it again in an open-access survey:  "the most striking thing in the eyes of many is that Muhammad is not mentioned on any media until the 680s."¹³

The silence is real, and it is not for lack of documents

The force of the argument comes from what the record does contain.  The Arabs who conquered the Near East left dated material almost immediately, and they named their own leaders in it.  Here is what survives from before AH 72, in Hoyland's catalogue of dated Muslim writings:¹

Read that table the way a sceptic reads it.  Muʿawiya is named in a Greek bath dedication during his own reign, to the day.  ʿUmar's death is recorded in a private rock graffito in the very year it happened.  The seventh-century Near East was perfectly capable of leaving fast, dated, documentary traces of named individuals.  So a fifty-four-year gap before the founder of the religion appears in his own community's Arabic is not what a thin record looks like.  It is what an absence looks like.

The epigraphic pattern points the same way.  Frédéric Imbert, working the Arabic graffiti corpus, headed a section of his 2011 survey with the question "Un islam primitif sans prophète ?" — "A primitive Islam without a prophet?"  In Nevo's Negev corpus of 435 graffiti, Muhammad is named 17 times, under four per cent, and Moses is named more often than he is.¹⁴  In 677 texts from Arabia dated or datable to the first two Islamic centuries, the prophet appears 64 times — twelve of those in the first century and fifty-two in the second.¹⁴  The earliest dated graffito in Arabia naming him is from 121/738-39, and it names him beside Abraham.¹⁴  Imbert's own conclusion about timing is that Muhammad enters the epigraphic record around AH 65, under ʿAbd al-Malik, and that "his appearance in the graffiti seems to have been rather sudden," first in official media and only then in private writing.¹⁴

Nevo and Koren: no Muhammad before 690

Yehuda Nevo was an archaeologist who surveyed and published Arabic rock inscriptions in the Negev;  Crossroads to Islam was finished by Judith Koren after his death in 1992.  Their thesis is stated bluntly:

Muhammad is not a historical figure, and his official biography is a product of the age in which it was written (the 2nd century A.H.).  Muhammad entered the official religion only ca. 71/690, and the very few passing references to him in earlier literary sources should be regarded as later interpolations by copiers who knew the Traditional Account.  It is much more difficult to explain why, if he existed and played the central role accorded him in the Traditional Account, there are no references to him before 71/690 not only in the popular inscriptions but also where they should have been obligatory: on the coins and in the official pronouncements of the Arab State¹⁵

Their epigraphic contribution is real and should not be dismissed.  Nevo sorted his Negev inscriptions into classes by creed, and named the earliest class "Indeterminate Monotheism" — private supplications by ordinary people whose religious content "lacks distinguishing features which would enable one to class it with a known religion."¹⁵  Then a later class adds Muhammad, and a later class still adds recognisably Islamic theology.  That is a stratigraphy of belief, drawn from the writing of ordinary people rather than from the state.

And they know they need an argument for reading silence as evidence, so they give one:

If one's universe of discourse is defined as how an event took place, not whether it did, accounts which do not mention that event are worth very little. … But if one is trying to ascertain whether the event took place at all, the situation is different.  One unshakeable contemporary account would settle the question, and the fact that there is none, if demonstrable, becomes "positive" evidence.¹⁵

Ohlig, Popp, and Luxenberg: "Muhammad" as a title for Jesus

The Saarbrücken circle around Karl-Heinz Ohlig, which publishes under the name Inarah, pushes further.  Their reconstruction is that the Arab rulers of the seventh century were Arabian Christians, that the era beginning in 622 was originally a Christian Arab era, and that the figure of Muhammad separated out from the figure of Jesus only later.  Ohlig:  "long before the appearance of the idea of a Hijra, there was an Arabian-Christian reckoning of time, which began with the year 622 and was only later 'converted' to a Muslim meaning."¹⁶  Volker Popp, who supplies the coin and inscription argument, writes that the hijra "is not historical.  In reality, though, a new era did begin: the period of the self-government of Christian Arabs."¹⁶

The philological core is Christoph Luxenberg's, and it is the sharpest single move in the whole sceptical literature.  The Dome of the Rock inscription of 72/691-92 contains the Arabic muhammadun ʿabdu llahi wa-rasuluhu.  Everyone reads this as "Muhammad is the servant of God and His messenger."  Luxenberg reads muhammad not as a name but as what it is by grammar — a passive participle, "praised" or "the one to be praised" — and takes the clause as a Christological doxology:  "Praised be the servant of God and his messenger."¹⁶  His excursus heading states the thesis in his own capitals:  the gerundival participle muhammad used "AS A METAPHORIC APPELLATION FOR JESUS CHRIST AMONG THE PRE-ISLAMIC CHRISTIAN ARABS."¹⁶

The observation underneath this is a genuine datum, and any account of the Dome of the Rock has to deal with it:  the earliest securely dated occurrence of muhammad sits inside an aggressively anti-Trinitarian text that is about Jesus.  Five of the Quranic passages quoted on the inner arcade concern Christ.

Robert Spencer and Jay Smith: the popular version

Robert Spencer's Did Muhammad Exist? (2012;  revised and expanded, 2021) is the book most English-speaking readers meet.  Its argument is that the traditional account dissolves on inspection:  "The more one looks at the origins of Islam, the less one sees."⁸  His summary list in the final chapter includes these items, which are worth quoting because several of them are simply true:⁸

• "No record of Muhammad's reported death in 632 appears until more than a century after that date."
• "The early accounts written by the people the Arabs conquered never mention Islam, Muhammad, or the Qur'an. They call the conquerors 'Ishmaelites,' 'Saracens,' 'Muhajirun,' and 'Hagarians,' but never 'Muslims.'"
• "The Arab conquerors, in their coins and inscriptions, don't mention Islam or the Qur'an for the first six decades of their conquests."

Spencer's positive reconstruction is that the Arab empire, like every empire of its day, needed a political theology, and built one.  The rulers began with "a monotheistic religion centered around Abraham and Ishmael," and out of that, around the time of ʿAbd al-Malik, "the nascent religion began to take shape as an entity in its own right — a forthrightly, even defiantly, Arabic one.  The specific features that emerged revolved around the person of the 'praised one,' Muhammad, an Arabian prophet who may have lived decades before and whose words and works were already shrouded in the mists of history."⁸

Note that last clause carefully, because it is the most honest sentence in the sceptical literature and it concedes the point:  Spencer's own reconstruction allows for a real Arabian prophet.  His claim is about the Muhammad of Islamic tradition, not necessarily about a man.

Jay Smith, of the Pfander Centre for Apologetics, argues the harder version in public.  His statement of it, from his own channel's description of the 2020 debate, is that "the Muhammad of the later Traditions did not exist because there is simply is no historical evidence for him, nor his city Mecca, nor his book the Qur'an, nor his religion Islam, or its people the Muslims at all at least before 690 AD, a full 60 years when he supposedly would have died."¹⁷  In his foreword to Spencer's revised edition he turns the argument from silence around:  "the onus is no longer on us.  It is now on them to convince us that indeed their Muhammad is historical;  otherwise, it is they, and not us, who now 'argue from silence.'"⁸

And the methodological objection behind all of it

John Wansbrough, whose dating of the Quran is now dead, made a methodological point that is not dead.  The Islamic literary sources cannot, he said, "provide … an account of the 'Islamic' community during the 150 years or so between the first Arab conquests and the appearance, with the sīra-maghāzī narratives, of the earliest Islamic literature."¹⁸  Whatever one thinks of his conclusions, that gap is real and every scholar in the field concedes it.

Put at its strongest, then, the sceptical case is this.  The only detailed account of Muhammad comes from books written five to eight generations after him, by a community with every reason to construct a founder.  Strip those books away and the Arabic record of the seventh century contains a conquering state, a God, and a series of named commanders — and no prophet, for fifty-four years, in a record that names commanders within months.  Then, in the middle of a civil war, the prophet appears everywhere at once:  on rebel coins, on the caliph's coins, on his monument in Jerusalem.  That is what a state adopting an ideology looks like.  It is not obviously what a state remembering its founder looks like.

That argument deserves an answer, not a sneer.  Here is the answer.

The witnesses from outside, in order

The people the Arabs conquered wrote things down.  They wrote in Greek, Syriac and Armenian, they disliked the conquerors, and their writings survive independently of anything Muslims preserved.  Taken in date order they are the decisive evidence, and they are the reason no historian of the period doubts that Muhammad lived.

One caution before the list.  These are outsiders, and most of them are reporting what they heard.  Sinai concedes the point:  "it is likely that such references to Muhammad in non-Islamic sources are ultimately reliant on statements made by the Muslims themselves."  His next sentence is the one that matters:  "Nevertheless, the fact remains that the Arab conquerors must have been considering themselves to be followers of Muhammad already in about 640 ce."⁶  Hearsay about a man's teaching is still evidence that the man was there to be talked about.

July 634 — a Greek tract names a prophet, two years after the death date

The Doctrina Iacobi nuper baptizati, "The Teaching of Jacob, the Newly Baptized," is a Christian anti-Jewish dialogue set in Carthage among Jews forcibly baptised on Pentecost 632 by order of the emperor Heraclius.  Its own internal date is explicit:  Jacob leaves Carthage "on the thirteenth of July in the seventh indiction," which is 634.¹  Hoyland argues the tract was written then or very soon after, on three grounds:  the forced baptism of 632 that occasions the whole work becomes "both irrelevant and inexplicable" a decade later;  the news of the Arabs is relayed as rumour, which nobody could plausibly do for long;  and by December 634 the patriarch of Jerusalem is already writing about the Saracen sword as a present fact.¹  Shoemaker reports "a strong scholarly consensus" for the early date.²

The passage, in Hoyland's translation:

And they were saying that the prophet had appeared, coming with the Saracens, and that he was proclaiming the advent of the anointed one, the Christ who was to come.  I, having arrived at Sykamina, stopped by a certain old man well-versed in the scriptures, and I said to him: "What can you tell me about the prophet who has appeared with the Saracens?"  He replied, groaning deeply: "He is false, for the prophets do not come armed with a sword." … So I, Abraham, inquired and heard from those who had met him that there was no truth to be found in the so-called prophet, only the shedding of men's blood.  He says also that he has the keys of paradise, which is incredible.¹

Four things in that paragraph are testable against Islam, and Anthony lists them:  the prophecy begins among the Saracens;  warfare is integral to it;  it affirms the coming of the Anointed One;  and the prophet claims the keys of paradise.  Anthony judges the first three "as much confirmation as one might expect in the Qurʾan itself."³  Shoemaker calls the text "a very high-quality witness … contemporary with the events that it describes," and notes that it "purports to derive its information from an eyewitness source in Palestine who had confirmed what he writes with followers of Muhammad who had met him."²

Two honest problems.  First, the prophet is not named.  Could he be somebody else?  Anthony's answer is that the Islamic tradition itself names several rival Arabian prophets of Muhammad's lifetime, and "not one of these contemporary claimants to prophecy operated near Palestine, nor did they share Muḥammad's irredentist ambitions within Arabia or seek to lead conquests outside of it."³  That is a strong argument.  It is not a proof.  Spencer concedes the same point from the other side:  "It is difficult to imagine that it could refer to anyone else, as prophets who wielded the sword of conquest in the Holy Land — and armies acting on the inspiration of such prophets — were not thick on the ground in the 630s."⁸

Second, the text has the prophet coming with the Saracens, present tense, in 634 — two years after he is supposed to have died.  Shoemaker takes this seriously and builds a book on it:  Muhammad may have died later than 632 and led the invasion himself.²  That is a minority position and this article does not adopt it.  But notice which way it cuts.  If Shoemaker is right, the earliest external evidence is about a living man, not a legend.

Late 634 — Sophronius fixes the invasion to the month

Sophronius, patriarch of Jerusalem, never names Muhammad.  What he supplies is an eyewitness date.  In his Christmas homily of 25 December 634 he explains why his congregation cannot walk the few miles to Bethlehem:  they are "restrained from rushing over there, not seeing a twirling and flaming sword but the sword of the savage and barbaric Saracens, which, filled with every diabolical cruelty, striking fear and bringing murder to light, keeps us banished from this blessed vision."²

That is not evidence for Muhammad.  It is evidence that something enormous and Arab was happening in Palestine in exactly the months the Doctrina describes, reported from inside Jerusalem by a named man on a fixed date.  It anchors the Doctrina in real events.

c. 637 — the name, on the flyleaf of a Gospel book

British Library Add. 14,461 is a sixth-century Syriac Gospel manuscript of Matthew and Mark.  On its front flyleaf someone wrote twenty-three lines of notes about the fighting going on around him.  The lines are now badly rubbed.  They name Muhammad twice, and they date the great battle at Gabitha — the engagement Arabic sources call Yarmuk — to 20 August in the Seleucid year 947, which is 636.  Because the last surviving line begins a notice for the following year, the note is generally dated to 637.²

… Muhammad … the priest Mar Elias … and they came … And in January [the people of] Homs made an agreement in exchange for their lives, and many towns were destroyed in the slaughter by [the Nomads of] Muhammad, and many people were killed and captives [were taken] from Galilee all the way to Beth … And on the twentieth of August in the year [nine hundred and forty-]seven there assembled in Gabitha … the Romans and many people were killed, from the Romans about fifty thousand.²

Anthony:  "The earliest extant document to mention Muḥammad tout court by name … Written a mere five years after Muḥammad's death and, therefore, by a contemporary of the Arabian prophet."³  Shoemaker adds a point that is easy to miss:  this scrap "is even earlier than the Qurʾan, at least as a written document, by any estimation of the Qurʾan's origins."²  Andrew Palmer reads it as a parish priest's private jottings made as the news came in.

The honest caveats, and they are real.  The page is rubbed, and Anthony reproduces a photograph with arrows pointing at "faintly legible traces" of the name.  The spelling is MWḤMD rather than the later standard MḤMṬ — unusual, though paralleled in an eighth-century Syriac disputation text.  And the notice gives Muhammad no religious role whatever.  Anthony states this plainly:  "Absent is any mention of any notion of prophetic authority undergirding his leadership or, for that matter, any religious inspiration motivating the conquests."³  He is a war-leader here, and nothing more.

c. 640 — Thomas the Presbyter, and a date that checks three ways

Thomas was a Syriac priest;  his chronicle survives in an eighth-century manuscript, and Andrew Palmer has argued that seven texts in it form one covert world chronicle by a single compiler, who inserts himself into the record as the brother of Simon, doorkeeper of the monastery of Qedar.  The chronicle is dated to about 640 because it mentions Heraclius reigning for thirty years and records nothing later;  Heraclius died in February 641.¹

In the year 945, indiction 7, on Friday 4 February (634) at the ninth hour, there was a battle between the Romans and the Arabs of Muḥammad (tayyāye d-Mḥmt) in Palestine twelve miles east of Gaza.  The Romans fled, leaving behind the patrician bryrdn, whom the Arabs killed.  Some 4000 poor villagers of Palestine were killed there, Christians, Jews and Samaritans.  The Arabs ravaged the whole region.¹

Now check the date, which is the point of the passage.  The Seleucid year 945 runs from autumn 633 to autumn 634, so 4 February falls inside it.  AD 634 is the seventh year of the Roman fifteen-year indiction cycle, which the chronicler states independently.  And 4 February 634 in the Julian calendar really was a Friday.  A chronicler inventing a date at random gets any one of these right about one time in seven to one time in fifteen.  Getting all three right is what an actual record looks like.  Hoyland:  "its very precise dating inspires confidence that it ultimately derives from first-hand knowledge."¹  Muslim historians independently place a battle at Dathin, near Gaza, in the spring of 634.

Two cautions on this passage.  The casualty number is unstable:  the Latin translation of the Syriac printed in 1904 reads "about forty thousand," while Hoyland, Shoemaker and Michael Penn all render four thousand.¹⁹  Do not quote a number from it without saying which reading is followed.  And Sean Anthony's 2020 book misprints the year as 954, which would be 642-43;  945 is the manuscript reading and the one every other translation gives.

Spencer's answer to Thomas is worth stating because it is the best available.  He notes that the Syriac writes MḤMṬ, with a final t, not MḤMD;  that "there is some indication that these writings were revised in the middle of the eighth century";  and that even granting the translation, "we are still a considerable distance from the prophet of Islam."⁸  The last point is correct and this article agrees with it.  The first two are answers to a question nobody asked:  the issue is whether a man called Muhammad led these Arabs in 634, and a chronicle written six years later saying that they were "the Arabs of Muhammad" is exactly the evidence that he did.

655 — an Armenian bishop gives his preaching at length, and names his source

The Armenian chronicle conventionally called Sebeos is anonymous;  the old attribution to a Bishop Sebeos was disproved, and the name survives only as shorthand.  James Howard-Johnston, who produced the standard edition with Robert Thomson, dates the relevant section to spring or early summer 655, with some later material added around 661.⁴  Shoemaker traces the Muhammad material to a source of the 640s.

At that time a certain man from among those same sons of Ismael whose name was Mahmet, a merchant, as if by God's command appeared to them as a preacher [and] the path of truth.  He taught them to recognize the God of Abraham, especially because he was learned and informed in the history of Moses. … So Mahmet legislated for them: not to eat carrion, not to drink wine, not to speak falsely, and not to engage in fornication.  He said: 'With an oath God promised this land to Abraham and his seed after him for ever. … But now you are the sons of Abraham, and God is accomplishing his promise to Abraham and his seed for you.  Love sincerely only the God of Abraham, and go and seize your land which God gave to your father Abraham.  No one will be able to resist you in battle, because God is with you.'⁴

Then the chronicler says where he got it:  "We heard this from men who had been taken as captives from Arabia to Khuzistan.  And having been eyewitnesses of these things themselves, they told us this account."²

Two features of this passage are hard to explain on any theory of invention.  The four prohibitions — carrion, wine, falsehood, fornication — are every one of them Quranic prohibitions, reported by an Armenian churchman who had never seen a Quran and whose informants were prisoners of war.  And Muhammad is a merchant, which is what Jacob of Edessa independently says in Syriac forty years later, and which is not what the hadith tradition emphasises;  the hadith prefer to make him a shepherd.³  Two Christian writers in different languages agreeing on a biographical detail that the Muslim tradition does not push is the signature of independent information.

Sebeos is checkable, and he checks out where he can be checked.  On the Persian sack of Jerusalem in 614 he agrees with independent sources on the May date, the nineteen-day siege, the undermining of the walls, and the removal of the Cross.  He also makes real errors:  he conflates the Persian assaults of 615 and 626, and dates the collapse of one peace treaty to two different years.¹  And there is one thing he does not say, which the article should not hide:  he never mentions Muhammad bringing a book.¹

The 660s — Iraq, and a chronicler who does not know what the Arabs' shrine is

The Khuzistan Chronicle, written in Syriac by an East Syrian Christian around 660, describes the conquest of Persia:  "Then God raised up against them the sons of Ishmael, [numerous] as the sand on the sea shore, whose leader (mdabbrānā) was Muḥammad (mḥmd).  Neither walls nor gates, armour or shield, withstood them, and they gained control over the entire land of the Persians."¹

The same chronicler turns to Arabian geography and reveals how little he knows — which is itself valuable, because it shows he is not copying a Muslim account:

Regarding the dome of Abraham, we have been unable to discover what it is except that, because the blessed Abraham grew rich in property and wanted to get away from the envy of the Canaanites, he chose to live in the distant and spacious parts of the desert. … Indeed, it was no new thing for the Arabs to worship there, but goes back to antiquity, to their early days, in that they show honour to the father of the head of their people.¹

Hoyland's reading of this is the model for how to use these sources:  the chronicler "has evidently heard that the Arabs have a sanctuary which has some connection with Abraham.  Wanting to know more, he turns to Genesis and pieces extracts together until he has constructed an explanation that satisfies him."¹  So the passage proves that in the 660s Christians in Iraq knew the conquerors had an Abrahamic desert sanctuary and did not know what it was.  It does not confirm Mecca.  In fact the chronicler's list of Arabian places names Hasor, Medina, Dumat Jandal, the land of the Hagarites, Hatta, Mazon, Yamama, Tawf and Hira — and Mecca is absent.  That is a real datum and it belongs in the article, not in a footnote.

The Maronite Chronicle, from about 664, reports that in 660-61 "many Arabs gathered at Jerusalem and made Muʿāwiya king," that he struck coinage "but it was not accepted, because there was no cross on it," and that "he set his throne in Damascus and refused to go to the throne of Muhammad."¹  A "throne of Muhammad" somewhere else — Medina — which the new ruler declined to occupy is not the sort of detail an inventor supplies.

687 — John bar Penkaye, who knows what the Arabs enforce

John bar Penkaye finished his Book of Main Points in 687 or 688;  Hoyland dates it from the famine and plague he describes as happening "today" in "the year 67 of the rule of the Arabs."¹  On Muhammad:

And at their beginning they upheld the tradition [mašlmānūtā] of Muhammad, who was their teacher [tārʾā], so that they would confer a death sentence on anyone who was seen to be audacious against his laws.²

Hoyland is careful about the vocabulary and the article follows him:  "The term 'tradition' (mashlmānūtā) implies something handed down, but one doubts that a fixed corpus of rulings from Muḥammad is meant.  Most likely John is simply relaying the message given out by the Muslims themselves, that they adhere to and enforce the example of their Prophet."¹

Why this short passage matters more than its length suggests is that John can be independently checked, and he passes.  He reports that the rebel Mukhtar freed the slaves of the Kufan Arabs and put them in his armies.  Ninth-century Muslim historians report the same thing.  Hoyland:

This conforms to no polemical or stereotypical formula, and on internal grounds one cannot argue that John's history was revised at a later date.  The only possible explanation is that John is recording what had reached him from Muslim informants, probably via their Christian slaves, and that Muslims preserved this account until the ninth century.¹

That is the crux, and it generalises.  Where a seventh-century Christian and a ninth-century Muslim agree on a detail neither had reason to invent and neither could have copied from the other, the ordinary explanation is that both are reporting something that happened.

c. 692 — Jacob of Edessa, and the passage that cuts both ways

Jacob was bishop of Edessa and the leading Syriac scholar of his generation.  His chronicle, finished around 692, puts under the 349th Olympiad — about 617-18 — the note that "Muhammad travelled to conduct trade in the provinces of Palestine, of the Arabias, of Phoenicia, and of the inhabitants of Tyre," and dates the beginning of the Arab dominion, with Muhammad as its first king, to 622.³  That is the hijra year, arrived at by a Syriac bishop who never mentions the hijra.

Jacob also wrote the single strongest documentary support the sceptics have, and honesty requires printing it.  Answering a question about which way Jews pray, he writes:

the Jews do not worship toward the south, nor do the Muhājirūn [mhaggrāyē].  For as I saw with my own eyes and write to you now, the Jews who live in Egypt and in fact the Muhājirūn there also prayed toward the east, and both peoples still do: the Jews toward Jerusalem and the Muhājirūn toward the Kaʿba.  And those Jews who are south of Jerusalem pray to the north, and those in the land of Babel and in Ḥira [Kufa] and Basra pray to the west.  And the Muhājirūn there also pray to the west toward the Kaʿba …²

From Egypt, Mecca is south-east;  from Iraq it is south-west.  Jacob says east and west.  Shoemaker draws the strong conclusion:  "the Kaʿba revered by Muhammad's Near Eastern followers was not yet identified with Mecca but seems to have been located instead somewhere in Palestine."²  Hoyland, working the same problem, reaches a different verdict:  early prayer directions vary because working out the direction was genuinely hard without instruments, and Muslim legal sources say so themselves.  His conclusion:  "Early governors probably applied different principles according to their own views, many deeming prayer in the general direction of the Kaʿba as sufficient.  From the time of ʿAbd al-Malik and Walīd onwards there was a move to fix the qibla in the exact direction of the Kaʿba itself."¹

This article does not decide between them.  It notes what both readings share:  even on Shoemaker's reading, Jacob's testimony presupposes Muhammad and his emigrants as real people, and Jacob's own chronicle names Muhammad, gives his trade, and dates his rule from 622.  Moving the sanctuary is not the same operation as deleting the prophet.

What the outsiders establish, taken together

Hoyland's summary of what can be reconstructed about early Islam from non-Muslim writers alone, without using a single Muslim source, is the single most useful paragraph in his book:

Non-Muslim writers of the first century AH attest that it was strictly monotheistic (Sebeos, John bar Penkaye, Anastasius of Sinai) and iconoclastic (anti-Jewish polemicists, Germanus); that its adherents had a sanctuary, their "House of God" (Bar Penkaye), of Abrahamic association (Chronicler of Khuzistan, Jacob of Edessa), called the Ka'ba (Jacob of Edessa), towards which they prayed (Jacob of Edessa) and at which they sacrificed (Anastasius of Sinai) and reverenced a stone (Anastasius of Sinai, Germanus); and also that they followed Muḥammad (Thomas the Presbyter, Sebeos, Chronicler of Khuzistan), who was their "guide" and "instructor" (Bar Penkaye), whose "tradition" and "laws" they fiercely upheld (Bar Penkaye) and who prescribed for them abstinence from carrion, wine, falsehood and fornication (Sebeos).¹

His conclusion from it:  "It is thus evident that the early Muslims did adhere to a cult that had definite practices and beliefs and was clearly distinct from other currently existing faiths."¹  He directs that explicitly against Wansbrough's claim that the non-Muslim testimony is too vague to identify a community.

The documents, the stones, and the coins

The literary witnesses are outsiders.  The documentary evidence is different in kind:  it is what the conquerors themselves wrote, on papyrus and on rock and on metal, at dates they put on it themselves.  It is thinner than the literary evidence and it says less.  What it does say is load-bearing.

A calendar counted from an emigration

PERF 558, in Vienna, is a bilingual Greek-Arabic receipt for sixty-five sheep delivered to an Arab commander in Egypt.  The Arabic names the scribe and gives the date as Jumada I of the year 22;  the Greek gives the equivalent Egyptian month and indiction.  That is April 643 — eleven years after the traditional death date.³

Twenty-two years into what epoch?  Anthony:  "The most compelling answer to this question remains, 'twenty years after the hijrah of Muḥammad,' a hijrah that becomes the archetype for the subsequent hijrahs of the Arabian tribesmen to the conquered territories.  Indeed, at this juncture this answer remains the only feasible one — particularly when read in light of the pivotal, conceptual role of hijrah in the qurʾanic corpus, which renders this inference virtually irrefutable."³  Nobody in the seventh century ever writes down what event started the era.  Every alternative explanation offered has, in the judgement of the specialist literature, failed.

The second feature of the same papyrus is the Greek word on the back:  the conquerors are called magaritai, a coinage from Arabic muhajirun, "emigrants" — people who have made a hijra.³  A movement that calls itself the Emigrants and dates its documents from an emigration is not a movement without a story.

The next year, 23 / 643-44, a man scratched his name on a rock at al-Muthallath near Yanbuʿ in the Hijaz:  "Salma wrote [this in] twenty-three."  He too used the Hijri era, and he did it in a private graffito with nobody watching.³

The Zuhayr inscription: a private graffito that gets a caliph's death right

In 1999 ʿAli ibn Ibrahim al-Ghabban found, at Qaʿ al-Muʿtadil in north-west Arabia, a rock inscription that Hoyland renders:

"I am Zuhayr mawla (client) of Ibnat Shayba" (note the diacritical marks on the letters n and z), is next to another by the same person which he dates to "the time when ʿUmar died, the year 24," that is, 644 CE, making these the earliest Arabic inscriptions (but not the earliest Arabic texts, since there are papyri from the year AH 20/640 CE).²⁰

Anthony gives the whole sentence:  "In the name of God.  I, Zuhayr, wrote (this) at the time ʿUmar died, the year four and twenty."³

Consider what this small object does.  The caliph ʿUmar was stabbed on the last night of the twelfth month of AH 23 and buried on the first day of AH 24.  A stonecutter with no access to any chronicle, writing for himself, dates his graffito by the caliph's death — and lands within about a day of the date the Muslim tradition independently preserves and did not write down for three centuries.  You cannot fabricate that convergence backwards.  It also shows a functioning Hijri calendar in private use twelve years after Muhammad's death, and a developed system of dots distinguishing Arabic letters, which incidentally undercuts the premise that seventh-century Arabic script was too crude to fix a reading.

Two caveats, both from the epigraphist Frédéric Imbert, who is friendly to the sceptics.  He argues the invocation of God above the line is a later addition, because inscriptions beginning with the writer's name are common and are never preceded by that formula.  And he frames the whole find as a problem rather than a proof:  "Ce graffito pose la question de l'absence de références religieuses seulement 12 ans après la mort du prophète Muḥammad" — this graffito raises the question of the absence of religious references only twelve years after the prophet's death.¹⁴  He is right that it does.  It is also true that the writer used the prophet's calendar to say so.

The coins: the name appears first on rebel money

The numismatic sequence is the sceptics' best material and it should be laid out exactly.

From AH 31 onward, Arab-Sasanian silver coins pour out of mints across Iran in large quantities.  They carry only "in the name of God," sometimes "God is my Lord" or similar.  The obverse keeps the Persian shah's portrait;  the reverse keeps the Zoroastrian fire altar.  No Muhammad.

Then, in AH 66 and 67 — 685 to 687 — two drachms from Bishapur carry the Arabic legend bism Allah Muhammad rasul Allah, "In the name of God, Muhammad is the messenger of God."¹  This is the earliest surviving occurrence of Muhammad's name in Arabic written by his own followers.  It is fifty-three or fifty-four years after 632.

And the coins are rebel coins.  Hoyland spends four pages identifying the governor who issued them and concludes he was ʿAbd al-Malik ibn ʿAbd Allah ibn ʿAmir, married to the sister of Ibn al-Zubayr, governing Fars for the Zubayrid counter-caliphate that held Mecca against the Umayyads during the second civil war.  His conclusion:  "If this reconstruction is correct, it would mean that the earliest attested Islamic profession comes from an opposition party."¹  Anthony reaches the same conclusion independently, and notes that these coins lack the usual companion phrase "there is no god but God."³

Then Hoyland draws the inference that answers the whole argument from silence:

The total lack of specifically Islamic declarations made by the Sufyanid line of Umayyads, the proliferation of them issued by the Marwanid branch and the religious causes espoused by the various opposition movements of the intervening civil war lead us to the conclusion that it was pressure from rebel factions that induced the Marwanids to proclaim Islam publicly as the ideological basis of the Arab state.¹

Chase Robinson's version, quoted by Anthony:  "the Marwanids seem to have learned a Zubayrid lesson — that … principal articles of belief should be proclaimed and disseminated publicly."³

One rule that must be kept:  never cite a coin bearing the bare name "Muhammad" without the honorific as evidence for the Prophet.  Hoyland warns that a few such coins exist and that the name is "almost certainly a reference to a governor," probably Muhammad ibn Marwan.¹

The Dome of the Rock, 72 AH / 691-92

ʿAbd al-Malik's building in Jerusalem still stands, and its mosaic inscription still carries its own date.  "Muhammad is the messenger of God" occurs six times in it.³  One unit reads:

In the name of God the Merciful the Compassionate.  There is no god but God alone.  He has no associate.  To Him belongs dominion, and to Him belongs praise.  He gives life, He causes death, and He has power over all things.  Muḥammad is the messenger of God, may God bless him and accept his intercession on the Day of the Resurrection on behalf of his community.³

The inner face carries a chain of Quranic passages about Christ, quoted in wording that matches the received text:  Quran 4:171 and 4:172, Quran 19:33-36, Quran 3:19.¹  Two things follow.  This is the earliest dated monumental use of the Muslim profession of faith.  And a substantial body of Quranic text, in fixed wording, was in official circulation by 691-92.

The building is also, as the Inarah circle correctly observes, saturated with polemic about Jesus.  That is a fact to be explained, and the ordinary explanation is not strained:  ʿAbd al-Malik built a monument in a Christian city, facing the Christian holy places, and filled it with the Quran's own arguments against Christian doctrine.  Hoyland's note is the useful one:  "one should note not just the content, but the very fact that ʿAbd al-Malik presumed to dictate doctrine, to Christians as well as Muslims."¹

By 87/706 the formula is complete.  The mosque of Damascus carries:  "Our Lord is God alone, our religion is Islam and our prophet is Muhammad."¹

And one earlier item that may move the date

A tax document from Nessana in the Negev, P.Nessana 77, republished by Hoyland in 2015 and dated to the 60s AH — around 680 — carries the phrase "the pact of God and the pact of His Messenger."  If the reading holds, the earliest official invocation of the Messenger moves back from AH 66 to about AH 60.  Anthony cites it for that purpose.³  The 2015 article itself was not obtained for this project, so the article treats this as a lead rather than a load-bearing fact.

How many years? State the interval plainly

By the traditional reckoning Muhammad died on 8 June 632.  Here is the distance from that date to each kind of independent attestation, with no rounding in our favour:

Two years to the prophet.  Five years to the name.  Eleven years to a calendar counted from his emigration.  Fifty-four years to the name written by his own followers in their own language.

Every one of those numbers is in the article because leaving out the last one would be dishonest, and leaving out the first one is what the sceptical case does.

Why is the Arabic silent for fifty-four years?

This is the question the sceptics are right to press, and "the record is thin" is not an answer, because the record is not thin about ʿUmar or Muʿawiya.  Three things are needed for a real answer, and all three are available.

First:  the silence is about official Arabic media, not about the seventh century.  The literary attestation of Muhammad, in Greek, Syriac and Armenian, runs continuously from 634 to 687 and long predates the coins.  A movement that invented its prophet in the 680s would have to explain why Christians in Palestine, Mesopotamia, Armenia and Iraq had been naming him for fifty years — and, worse, why some of them got details right that the Muslim tradition itself does not stress, like the merchant.

Second:  early Islamic official media were not the place where religion got proclaimed, until somebody made them the place.  Hoyland's answer to Nevo and Koren is exactly this:  "Much has been made of this apparent absence of Muhammad's name from early Islamic official state documents, but one should remember that, apart from the fact that it is an argument from silence, such texts were not intended as historical reports. … Moreover, quite a number of non-Muslim sources mention Muhammad by name in the course of these first decades of Muslim rule."²¹

Third — and this is the positive explanation, not merely a rebuttal — the change in the 680s has a known political cause.  The name enters state media at a precisely identifiable moment:  the second civil war, on the coins of the Zubayrid rebels who held Mecca against the Umayyads, and then on the coins and monuments of the Marwanid Umayyads who beat them.  That is Hoyland's reconstruction, quoted above, and Chase Robinson's, and Anthony's.  It also fits Fred Donner's independent argument that the movement's public identity hardened into a separate religion called Islam under precisely this caliph, ʿAbd al-Malik.

The sceptical reading of that same moment is that the prophet was created then.  The mainstream reading is that he was advertised then.  The two readings predict different things about the fifty years before, and the fifty years before are full of Christians writing his name.  That is what decides it.

Compared with other founders

One reason the interval argument feels stronger than it is, is that most readers have no baseline.  Here is one.

Two things follow.  First, Muhammad's documentation is at the good end of the ancient range, not the bad end.  If two years to an external notice is not enough to establish that someone existed, then almost no pre-modern founder existed.  Second, the control rows are ʿUmar and Muʿawiya, and they are why the sceptical argument is not stupid.  This society did leave rapid documentary traces of named individuals.  So the fifty-four-year Arabic silence about Muhammad is a genuine phenomenon that needs an explanation, and it has one.

The strongest Muslim responses

On the question in this article's title, a Muslim reader and this article agree.  The disagreement is about the second half of the sceptical claim — whether the traditional biography survives — and about how much credit the Islamic sources deserve.  Here are the best serious arguments on that side, and where each one lands.

"You are still applying a double standard to our sources"

The objection:  the article accepts a rubbed flyleaf and a hostile Armenian chronicle as evidence, but treats thousands of carefully transmitted reports as literature.  If a Syriac priest's private note counts, why does an audited chain of named transmitters not count?

This is a fair challenge and it has real scholarship behind it.  Harald Motzki spent a career answering Joseph Schacht with Schacht's own tools, and got results.  On Schacht's model of law being back-projected onto the Prophet:  it "is a construct which is not tenable in this form.  Certainly there occurred many projections of opinions onto the Prophet and the ṣaḥāba, but this is a phenomenon which set in rather late."²³  And on the chronology:  Schacht placed the beginnings of Islamic law "too late by a good half to three quarters of a century."²³  Applying the same method to the Quran-collection reports, Motzki concluded that stories of a collection under Abu Bakr and an official edition under Uthman "were already in circulation towards the end of the 1st Islamic century."²⁴

The answer.  Two things are true at once, and the article holds both.  The Islamic material is much earlier than mid-twentieth-century scepticism allowed;  the sceptics have lost that argument and should stop making it.  But dating a report is not the same as verifying it, and Motzki himself says so in the same paragraph as his conclusion:  "We are not able to prove that the accounts on the history of the Qurʾān go back to eyewitnesses of the events which are alleged to have occurred.  We cannot be sure that things really happened as is reported in the traditions."²⁴  Anthony puts the same limit more sharply:  the method "cannot verify the historicity of a given tradition being transmitted. … An early tradition is neither necessarily a historically accurate tradition nor even a historical one."³

So the honest position is neither "the isnads prove it" nor "the isnads are worthless."  It is Hoyland's:  the central problem is not mass forgery but "transformation of a tradition's content in the course of transmission," which happens without anyone lying.²¹

"Your own scholars concede the sceptics are losing"

They do, and the article has already quoted them:  Berg on the consensus forming against the sceptics, Görke and Schoeler on the sceptical party having "shrunk to a small group."  Anthony's own summary is stronger still and a Muslim is entitled to quote it:

Based on these early cardinal sources, it seems beyond doubt that in the first half of the seventh century there emerged a law-giving claimant to prophecy from Arabia, a Saracen/Ishmaelite merchant from an Arabic-speaking tribe named Muḥammad, who claimed to culminate a long line of monotheistic Abrahamic prophets gifted with divine revelation.  Furthermore, we can deduce on a well-sourced evidentiary basis that this prophetic claimant, whose earliest followers regarded themselves as descendants of the biblical patriarch Abraham, formed a community in Western Arabia and became a ruler in Yathrib.³

The answer is to agree, and then to read the next two pages.  Anthony immediately adds that the seventh-century non-Muslim sources do not "yet know of the Qurʾan" — they speak of Muhammad's teachings, traditions or laws, never of a scriptural corpus.³  On the biography he refuses both extremes:  "This literature conveys too much in the way of raw data for it to be either ignored or categorically rejected. … But source minimalism is unwarranted."³  That is a much narrower result than "the tradition is vindicated."

"Even Patricia Crone gave up"

She did, twice, and it is the single most important fact about where this field has moved.  In 2006 she wrote:  "There is no doubt that Mohammed existed, occasional attempts to deny it notwithstanding."⁷  She named Nevo and Koren and rejected their inference:  "few would accept the implied premise that history has to be reconstructed on the sole basis of documentary evidence … The evidence that a prophet was active among the Arabs in the early decades of the 7th century, on the eve of the Arab conquest of the middle east, must be said to be exceptionally good."⁷  Shortly before her death she went further, on the Quran:  "we have a hard fact: the Qurʾān existed by the time when the tradition says it existed.  There is no longer any good reason to doubt that ʿUthmān set up a commission that produced a Qurʾān."²⁵  Her co-author Michael Cook, asked in 2021 about Hagarism:  "The short answer is, you know, I think we got an awful lot of it wrong."²⁶

The answer is to agree — and to note what she did not retract.  The same 2006 article keeps the Mecca problem at full strength:  "Not a single source outside Arabia mentions Mecca before the conquests, and not one displays any sign of recognition or tells us what was known about it when it appears in the sources thereafter. … In sum, we have no context for the prophet and his message."⁷  A Muslim who quotes Crone for the retraction has to take the rest of her too.

Common objections and responses

"Muhammad" is not a name, it is a title — "the praised one"

The observation is correct as grammar.  Muhammad is a passive participle of the root H-M-D and does mean "praised."  Luxenberg builds his reading of the Dome of the Rock on it, and Jay Smith uses it in debate.  Three things answer it.

First, "praised" is a perfectly normal Arabic personal name, then and now, as Ahmad and Mahmud also are;  meaning something is not the same as being a title.  Second, the flyleaf note of c. 637 and Thomas the Presbyter do not use the word in a religious formula at all.  They write "the nomads of Muhammad" and "the Arabs of Muhammad" — a possessive construction identifying a body of men by their leader, which is how you name a person, not how you praise a deity.  Third, Sebeos and Jacob of Edessa treat the word as the name of a man with a trade and a chronology:  a merchant, active around 617-18, whose dominion begins in 622.  Titles do not have careers.

Robert Spencer's version of the objection is more careful than most:  he says the flyleaf writer and Thomas "may also have meant to use the word 'Muhmd' or 'Mhmt' not as a proper name but as a title, 'the praised one' or 'the chosen one,' with no certain referent."⁸  "May" is doing all the work in that sentence, and the two texts that follow within twenty years — Sebeos and the Khuzistan chronicle — supply the referent.

The early Christian sources do not describe Islam, so they cannot be about Muhammad

Half true, and the true half is important.  Anthony is right that the earliest Syriac notices "exhibit not a shadow of cognizance that the conquests were inspired either by a prophet or, for that matter, any religious message at all."³  The first Syriac text to call Muhammad a prophet without ambiguity does not appear until the 690s.

But the objection proves the opposite of what it is offered for.  If Christian scribes had been retroactively inserting a prophet into their records, they would have inserted a prophet.  What they inserted was a war-leader with no religion attached — which is what a hostile outsider in 637 would actually have known.  A forger writes what the later tradition needs.  These texts do not.

The references to Muhammad in early Christian texts are later interpolations

This is Nevo and Koren's move, and it is the load-bearing move in their whole book:  "the very few passing references to him in earlier literary sources should be regarded as later interpolations by copiers who knew the Traditional Account."¹⁵

The problem is that it is unfalsifiable.  Any evidence that appears is reclassified as insertion.  Crone, who read their drafts and argued the other side, declined to follow them for exactly that reason.  And the specific texts resist it.  The flyleaf note is a private scrawl on a Gospel book, not a text anyone was curating.  Thomas the Presbyter's Muhammad notice is welded to a date that checks against three independent calendars — an interpolator would have to have inserted the name and got the Julian weekday of a battle right.  Sebeos names living informants.  John bar Penkaye's independently verifiable material about Mukhtar shows he had real Muslim informants, which is precisely the channel by which he would have learned the name.

But there is no seventh-century evidence for Mecca

Correct, and this article does not pretend otherwise.  Crone's formulation is the defensible one and should be used instead of the popular versions:  no source outside Arabia mentions Mecca before the conquests, the Quran describes its opponents farming wheat, grapes and olives, and Mecca cannot grow olives.⁷  The Khuzistan chronicler's list of Arabian places, written in the 660s, does not include Mecca.

Two things follow.  The Mecca problem is real and unsolved, and a Christian apologist may press it honestly.  But it is a problem about where, not about whether.  Establishing that a movement's sanctuary was somewhere other than the tradition says is not establishing that its founder did not exist — and Jacob of Edessa, the best witness against the Meccan geography, is also a witness for Muhammad's trade and dates.

The version of this objection that must be retired is the "qibla points to Petra" claim associated with Dan Gibson.  It is not a position in the field, and it is a weak plank in an argument that does not need it.

The absence of evidence is evidence of absence when evidence should be there

This is the strongest general form of the sceptical argument and Nevo and Koren state it well:  "One unshakeable contemporary account would settle the question, and the fact that there is none, if demonstrable, becomes 'positive' evidence."¹⁵

The principle is sound.  The application fails on its own terms, because the antecedent is false:  there are near-contemporary accounts, in Greek from 634 and Syriac from 637, and the argument only works if each of them is separately neutralised.  Note also what the principle demands of the sceptic.  If absence where evidence should be counts as positive evidence, then presence where evidence should be — a calendar in 643, a caliph's death date in a private graffito in 644, four Quranic prohibitions in an Armenian chronicle in 655 — counts too.

If the sira is unreliable, how can Christians use it against Muhammad?

They cannot, at least not in the way they usually do, and this deserves saying because it cuts against our own side.  The house rule for this series is that a report good enough to convict is good enough to acquit.  An apologist who calls the hadith worthless when defending the historicity question, and then quotes Bukhari on Muhammad's marriages, is arguing both sides of the same evidence.

The consistent position is the one this article takes:  state each report's status before using it — which collection, how early, what chain, who graded it — and let the weight follow.  David Wood's argument in the 2020 debate did exactly this, and did it against a fellow Christian:  he argued that the Muslim sources should be trusted about Muhammad because a tradition inventing a founder would not have invented the episodes that embarrass it.⁵  That argument has a known weakness — John Burton showed there was a doctrinal motive for inventing the most famous "embarrassing" episode, and Hoyland concludes the criterion fails as a general tool.²¹  But the instinct is right, and the alternative — using the sources when convenient and dismissing them when not — is not an argument at all.

Is this just Muslim consensus dressed up?

No, and it is worth being blunt about who holds this position.  The scholars quoted in this article are not Muslims.  Hoyland, Anthony, Crone, Cook, Donner, Sinai and Shoemaker are secular Western academics, several of them famous for scepticism about Islamic sources, and one of them — Shoemaker — is the field's leading current sceptic.  Chase Robinson, also a sceptic, writes:  "No historian familiar with the relevant evidence doubts that in the early seventh century many Arabs acknowledged a man named Muḥammad as a law-giving prophet."³  These people disagree with each other about nearly everything else in the field.  They do not disagree about this.

Shoemaker is the sharpest test, because he is the scholar with the most to gain from a late Muhammad.  In Creating the Qurʾan (2022) he argues at length that almost nothing the Islamic tradition says about Mecca and Yathrib in Muhammad's lifetime is usable — those accounts are "little more than pious fictions, with effectively no basis in any genuine historical memories."  And then, in the next sentence:  "Of course, to be clear, there is little doubt concerning a number of basic facts about Muhammad's prophetic career — for instance, that he existed, that he began a new monotheist religious movement in the Hijaz, the important influence of eschatology in shaping the movement, its connection to the Abrahamic tradition, and its emphasis on piety and conquest."²⁷  That is the sceptical position at its most sceptical, and it still includes "that he existed."

The argument inside Christian apologetics

This question is unusual in that the loudest disagreement is not between Christians and Muslims.  It is between Christians.

On 14 November 2020, in California, David Wood and Jay Smith debated "Did the Muhammad of Islam Exist?"  The event was hosted by the M2M Network.  Wood argued that he did exist and that the later traditions can be trusted about him;  Smith argued that he did not.⁵  Wood has taken the same affirmative side twice against Robert Spencer.  Two Christian apologists, arguing in front of a Christian audience, on opposite sides of a historical question — which is healthier than it sounds, and which is why the debate is worth naming.  Smith argued the same negative again in 2025 against Raymond Ibrahim, the translator of The Al Qaeda Reader, who took Wood's side;  PJ Media published the video on 22 August 2025.²⁸

The Berean Archive's judgement is that Wood is right and Smith is wrong, and that saying so matters more than scoring a point against Islam.

Smith's individual data points are often real and worth knowing.  The chronology of the canonical Quranic reading traditions is accurately reported.  The gap between the hadith compilers and their earliest surviving manuscripts is genuine.  The epigraphic silence before ʿAbd al-Malik is a real puzzle, as this article has said at length.  What does not follow is the conclusion.  And several of the supporting planks are contested or fringe:  the Petra qibla thesis, the identification of the seventh-century "Muhammad" with a Christian Lakhmid king, and the Lüling-Luxenberg reconstruction of the Quran from Syriac hymnody.

Some specific claims should be retired from Christian use altogether, on the same principle by which this site retires bad arguments for Jesus' historicity.

• "Muhammad may never have existed." Not held by anyone in the field. Retire it.
• "Wansbrough proved the Quran is a ninth-century compilation." Dead. Sinai calls the extreme version "by now an easily vanquished straw man."²⁹ Van Putten says the manuscripts show "this position in its most extreme form is to be discarded."³⁰
• "Luxenberg showed the Quran is a misread Syriac lectionary." No standing in the field. Crone, who had every reason to welcome it, wrote that Lüling's and Luxenberg's books "are open to so many scholarly objections (notably amateurism in Luxenberg's case) that they cannot be said to have done the field much good."⁷
• "We have better evidence for Jesus than for Muhammad." False as usually meant. The external evidence for Muhammad is earlier. The Christian case does not need this and is damaged by it.
• Quoting Hagarism without the retraction. Both authors have disowned its central thesis in their own published words. Quoting a 1977 book against two authors who spent decades walking it back is not scholarship.

Where this article's case is weak

• The earliest witnesses are hostile hearsay, and they give Muhammad no religion. None of them met him; Sinai's concession that they are probably repeating Muslim self-description is correct; and the flyleaf and Thomas the Presbyter know a war-leader, not a prophet. What survives all that is narrow but firm: by 640 the conquerors were describing themselves as followers of a man named Muhammad.
• The traditional death date is not secure. Several non-Muslim sources place Muhammad at the head of the invasion of Palestine, which began in 634. Sinai says the hypothesis that the tradition moved his death from after 634 back to 632 "can hardly be ruled out."⁶ Crone said the same. This does not help the mythicist — it makes the earliest evidence about a living man — but it does mean the interval numbers in this article's table carry a two-to-three-year uncertainty at the top.
• The Mecca problem is unsolved, and this article does not solve it.
• The biography is not established. Almost everything a reader knows about Muhammad — the cave, the first revelation, the migration, Badr, Uhud, the wives, the farewell sermon — comes from books written five to eight generations later. This article establishes that he lived and led. It does not establish that any given episode happened, and no honest treatment can.
• Fifty-four years of Arabic silence is a real anomaly, and the political explanation offered for it, though well evidenced, is a reconstruction rather than a documented fact.

What evidence would change the conclusion?

What would weaken the case that Muhammad existed:

• A demonstration that the Doctrina Iacobi is a later compilation. Paul Speck argued for an eighth-century date; his conclusions "have not been widely followed," but if they were sustained the earliest witness would drop out.³
• Physical or codicological proof that the name on the BL Add. 14,461 flyleaf is a later hand. The page is faint enough that this is a testable question, not a rhetorical one.
• A demonstration that Thomas the Presbyter's Muhammad notice was added in the eighth-century recension. Spencer gestures at this; proving it would be a serious blow.
• The identification of a different named seventh-century Arabian prophet who fits the Doctrina's description as well as Muhammad does.
• A convincing philological case that muhammad functions as a title in the flyleaf and in Thomas — one that survives the possessive constructions "the Arabs of Muhammad" and "the nomads of Muhammad."

What would strengthen it further:

• Publication of first-century Arabic inscriptions naming Muhammad from before AH 60. Anthony reports undated inscriptions from the Hisma region near Tabuk, by a man identifiable in the Arabic literary sources, carrying the full profession of faith, which their editor dates cautiously to the end of the first century AH — and Anthony's own forecast is that "my strong intuition is that the accuracy of some of these statements may soon be overturned by future discoveries."³
• Confirmation of the P.Nessana 77 reading, which would move the earliest official reference to "His Messenger" back to about AH 60.
• Documentary evidence from the Hijaz itself for the 620s and 630s. There is almost none, and it is the largest hole in the record.

What would change the picture of Mecca — the genuinely open question — would be an external seventh-century source that names it, or archaeology in the Hijaz on a scale nobody has yet been permitted to do.

The conclusion of this article is not fragile, but it is not unfalsifiable either, and the list above is what would break it.  That the list can be written at all is the difference between a historical argument and a slogan.

Sources