Read history

April 713: a treaty, not a last battle

In April 713 ʿAbd al-ʿAzīz ibn Mūsā issued a peace-and-surrender agreement to Theodemir, a ruler in southeastern Iberia whom Arabic sources call Tudmīr.

The listing joins that treaty to a supposed final Visigothic defeat at Segoyuela in Salamanca, on the other side of the peninsula. Segoyuela was asserted as a September 713 battle in an early-twentieth-century Cambridge history, while the municipality of San Miguel de Valero presents the story as tradition standing between reality and imagination.

A final defeat is hard to locate in 713 in any case: a Visigothic kingship under Ardo survived in Septimania until about 720.

What Theodemir kept, and what it cost

The agreement confirmed Theodemir's local authority and protected his people from killing, captivity and the separation of families, along with their churches and religious practice.

Seven towns submitted nominally to Muslim authority. Their inhabitants accepted an annual tax and obligations not to assist Muslim enemies or conceal information from them.

In al-ʿUḏrī's recension the annual charge is one dinar per free man and half that for a slave, plus four measures each of wheat, barley and vinegar, two of honey and one of oil.

How the text reaches us

The treaty survives only through much later Arabic works. Most researchers accept it as authentic conquest-period evidence, though Luis Molina has questioned aspects of its textual authenticity and historical setting.

The recensions disagree with each other. The earliest dates the document only to Rajab 94 AH; the precise 5 April comes from another recension, and among the town names only Orihuela, Mula, Lorca and Alicante are secure.

A later chronicle adds that women stood on Orihuela's walls with their hair arranged like beards — a motif also known from Lombard tradition.

Spring 714 on the upper Ebro

In spring 714 Mūsā ibn Nuṣayr and Ṭāriq ibn Ziyād conducted a joint northern campaign that a Cambridge scholarly synthesis credits with taking Zaragoza, the upper Ebro, León and Astorga.

Regional scholarship likewise dates Zaragoza's submission or conquest to that spring and attributes it to Mūsā. No exact day is known.

How the city changed hands is not established. The sources are sparse and mostly late, and it is unclear whether Zaragoza was stormed, capitulated or entered by negotiated submission.

Who was actually marching

The force was not simply Arab. Ṭāriq and many of his troops were Berbers, while Mūsā's later force included a substantial Arab contingent.

Philippe Sénac notes that the sources do not establish which western Ebro districts were subdued, so even Muslim control of La Rioja at this point is inference rather than a narrated itinerary.

The attribution of León and Astorga to the same campaign is accepted in major syntheses but cannot be reconstructed day by day from contemporary evidence.

Summoned to Damascus

Within the same year the caliph al-Walīd summoned both commanders to Damascus, ending their direct role in the Iberian campaign. Zaragoza's capture and the recall were separate developments inside 714, not one day's business.

Mūsā was governor of Ifrīqiya, based at Qayrawān — not the wali of al-Andalus with a capital at Seville.

After the recall his son ʿAbd al-ʿAzīz assumed the governorship of al-Andalus, approximately 714 to 716, and operated from Seville.

The year is probably wrong

The advance filed under 715 belongs a year earlier. The best-supported scholarly chronology places the northern campaign that took León in spring 714.

That account describes a joint campaign by Mūsā and Ṭāriq covering Zaragoza, the upper Ebro, León and Astorga. It identifies no separate advance led by Ṭāriq alone.

The Real Academia de la Historia places both the northward movement and the commanders' summons to Damascus in 714.

No documented march from La Rioja

Research on La Rioja stresses that the sources do not establish whether the western Ebro lands were fully subdued during the Zaragoza campaign, which makes the claimed La Rioja-to-León route conjectural.

Later accounts also disagree over whether Mūsā, Ṭāriq or their combined forces took particular northern cities.

No consulted scholarly source supports a distinct La Rioja-to-León march.

A walled city with no recorded siege

León was a fortified former Roman city, and it probably passed into Muslim-controlled territory during that campaign.

The surviving evidence narrates no dated battle or siege at its walls, and gives no reliable day or month beyond the spring 714 placement.

Occupation did not mean permanent, uncontested Muslim control of the Cantabrian north.

15 August 718: the withdrawal

On 15 August 718 the Umayyad besiegers withdrew from Constantinople after roughly thirteen months. Caliph Umar II had ordered the commander Maslama ibn ʿAbd al-Malik to retreat.

The land army began moving back through Anatolia while surviving ships carried other personnel away by sea. Leo III kept the capital.

The date comes from the Byzantine chronicle tradition and coincides with the Dormition feast; Arab narratives give no exact day, though both Theophanes and Nikephoros support the traditional ending.

Why it had become untenable

Famine, disease and a severe winter had worn the besiegers down, alongside Byzantine naval victories, defections from reinforcement fleets and the defeat of an overland relief army.

Bulgar pressure on the besiegers' rear contributed materially, but the sources do not securely date a single Bulgar assault to 15 August. They also disagree on when and why the Bulgars struck — Theophanes and al-Ṭabarī describe attacks on the Arab camp, while a Syriac account has foraging Arabs entering Bulgar territory.

The Bulgar commander is often named as Tervel, though Bulgarian chronological evidence may place his death before the siege.

The numbers, and who supplied them

Theophanes reports 22,000 Arabs killed in the fighting with the Bulgars, crediting people said to be well informed rather than a count he made. The figures of 30,000 and 32,000 come from the much later Western chroniclers Sigebert and Alberic.

During the naval retreat the sources report destructive storms and an extraordinary fiery hail that boiled the sea and sank ships. Reading that as a Thera eruption is an interpretation; the well-attested eruption of Thera came in 726.

Theophanes says five ships reached Syria — a chronicler's dramatic claim, not a verified count of surviving vessels.

What 718 can carry

A modern peer-reviewed reconstruction considers 718 the likeliest year in which Pelagius rebelled in the Asturian mountains against al-Ḥurr, the governor of al-Andalus. It does not securely date the later battle narrative to that same year.

Other proposed dates are 722 and 735–737. The 2019 study favours 718 for the revolt, while noting that 722 was a scholarly reconstruction whose arguments are not conclusive.

No contemporary eighth-century source supplies a day-by-day account of Pelagius's election or a securely dated Battle of Covadonga.

What the late chronicle narrates

The late-ninth-century Chronicle of Alfonso III says people in Asturias gathered and elected Pelagius as their princeps. It then narrates a separate punitive expedition commanded by Alqama, a confrontation at a cave on Mount Auseva, Alqama's death and Bishop Oppa's capture.

That chronicle was composed roughly 160 to 200 years after the alleged events and is heavily shaped by biblical and hagiographic models. Its total of 187,000 enemy troops splits into 124,000 dead and 63,000 fugitives — figures scholars trace to biblical numerology rather than military accounting.

In the same account, stones hurled at the cave reverse direction and strike the attackers.

The dispute among historians

A 2022 scholarly reassessment goes further, disputing whether the conventional Covadonga battle can be established independently at all. One recent reading assigns the Chronicle of 754's documented Pyrenean fighting to ʿAbd al-Malik ibn Qaṭan's 733–734 campaign instead.

Calling the episode the beginning of the Reconquista is a retrospective interpretive framework rather than a contemporary description.

Even Cangas de Onís, later treated as Pelagius's first capital, is absent from the primitive version of the Covadonga story.

140 days, then a return

The near-contemporary papal biography of Gregory II in the Liber Pontificalis says Lombards seized the fortress of Sutri by deceit and possessed it for 140 days; the standard year is 728.

Gregory II sent repeated letters and admonitions to King Liutprand, and supplied many gifts.

The source says the fortress had been stripped of its resources before the king restored and donated it to the apostles Peter and Paul.

Two verbs pulling different ways

The wording restituit atque donavit — restored and donated — is legally ambiguous. Some historians treat Sutri as an important step toward papal territorial power; others stress that it was a restoration of property already belonging to the Patrimonium Sancti Petri.

There is no treaty text behind it. The surviving evidence is a narrative in the Liber Pontificalis.

That passage names the fortress of Sutri specifically, and does not extend to Vetralla or a collection of unnamed Latian towns.

What 728 did not include

The primary account describes no treaty signed at Sutri and no meeting there between pope and king. Its account of Gregory personally meeting Liutprand near Rome belongs to a different campaign involving the dukes of Spoleto and Benevento.

Liutprand had not occupied all of the Exarchate of Ravenna. Treccani places a later capture of Ravenna under his nephew Hildeprand in the 730s, followed by a Byzantine recovery.

The Prosopography of the Byzantine Empire renders the occupation as 727/728 from the indiction evidence; no exact modern calendar day is supplied.

A march no source records

The surviving early narratives describe no September 732 march through Orléans, no Loire crossing, no camp at Ballan-Miré, no skirmishes between the Indre and Creuse, and no Umayyad withdrawal to the Vienne.

The near-contemporary Chronicle of 754 gives only a broad sequence: ʿAbd al-Raḥmān defeated and pursued Eudes, devastated forts and churches, intended to pillage Saint Martin's basilica at Tours, and then encountered Charles.

Oxford's current scholarly survey describes even the battle's date and precise location as uncertain.

What can be said about the approach

After defeating Eudes and ravaging through Aquitaine, ʿAbd al-Raḥmān's force moved toward the wealthy shrine of Saint Martin. Eudes sought Charles's assistance, and Charles assembled a Frankish army to intercept the advance before it reached Tours.

A modern source study places the eventual encounter on the road between Poitiers and Tours rather than securely at Ballan-Miré. Moussais and Ballan-Miré are modern proposals, not sites established by contemporary testimony.

The listed army strength of 30,000 has no support in the consulted early or scholarly sources.

Even the year is contested

The traditional battle date is October 732, but Oxford's 2024 bibliography notes that many scholars favour late 733 or 734 on Iberian chronological evidence.

In the reliably narrated sequence the armies faced one another for nearly seven days, then fought; ʿAbd al-Raḥmān was killed and the Umayyad army evacuated its camp during the night.

Raiding continued afterwards, and modern scholars debate the battle's macrohistorical importance.

Out of Córdoba after a reprimand

The campaign filed under 733 began with a rebuke. The near-contemporary Chronicle of 754 says Governor ʿAbd al-Malik ibn Qaṭan left Córdoba with the public army after being reprimanded for achieving no victory in Frankish territory, and set out to subdue or devastate the inhabitants of the Pyrenean heights.

Renewed raiding and expectations of booty or revenue appear to have driven it. The caliph Hishām ibn ʿAbd al-Malik is the prince whose dissatisfaction is understood to have prompted the move.

The chronicler emphasises failure, losses and retreat rather than conquest or colonization.

Lost in the narrow ground

The army was directed through narrow terrain, achieved nothing profitable and lost many warriors.

A small number of Christian defenders held the heights while the larger force returned to the plains by indirect routes.

The Chronicle establishes the Pyrenees but names no battlefield; identifying the defenders specifically as Basques rests on later Arabic traditions.

733 or 734

A peer-reviewed 2004 study connects the passage with Arabic notices of an expedition against the Basques in AH 115, a year running from 21 February 733 to 9 February 734.

A 2022 reassessment instead places the expedition in the summer of AH 115–116 and argues specifically for 734.

The consulted scholarship also distinguishes this operation from Covadonga, with which it is sometimes confused.

A fire moving east across the sea

In the annalistic year 745/746, by the account of the chronicler Theophanes, bubonic plague broke out in Sicily and Calabria and began moving east. He compared its spread to fire. The disease crossed the sea lanes to the Peloponnesian port of Monemvasia, then into mainland Greece and the nearby islands.

This was not the Black Death, which lay six centuries in the future. Modern scholars identify the outbreak as the plague of 746–747, most likely one of the final recurrences of the first plague pandemic that had struck the Mediterranean under Justinian two hundred years earlier.

What the sources can and cannot say

Theophanes wrote decades after the events and organized his chronicle by Byzantine world-years and indictions, which is why the outbreak's start converts to the awkward span 745/746 rather than a clean date. He records that in the following indiction the disease reached Constantinople, where it caused severe mortality during the reign of Constantine V.

The event's traditional framing — a third of Asia Minor's population dead — appears in no consulted source. The route Theophanes describes runs west to east through Greece, not through Asia Minor at all, and no consulted source supports the claim that one-third of Asia Minor's population died.

Plague reaches Constantinople in a later phase

The immediate consequence was that plague stood established in Monemvasia, mainland Greece and the islands, positioned on the sea routes to the capital. The Constantinople phase that followed — dated 747–748 in some reconstructions — caused severe mortality in the city.

The night of 25 Ramadan

On the night of 25 Ramadan 129 AH — converted in the published translation as 9 June 747 — Abu Muslim publicly unfurled two banners at Safidhanj, a village in the Merv oasis.

The banners had been sent by the Abbasid imam Ibrahim al-Imam and were named The Shadow and The Clouds. Abu Muslim raised them on separate lances and recited Qur'an 22:39, the verse granting permission to fight to people who had been wronged.

Abu Muslim, Sulayman ibn Kathir, their associates and responding villagers put on black clothing.

A village, not a battle in the city

This was an organized public proclamation in the Merv oasis rather than a battle inside Merv itself, and agents had been assigned to proclaim the movement elsewhere in Khurasan.

Ibrahim al-Imam had ordered Abu Muslim to expose the clandestine daʿwa, which had operated secretly for years. Abu Muslim moved while Umayyad authority in Khurasan was weakened by rivalry among the governor Nasr ibn Sayyar, al-Kirmani's faction and other rebels.

Groups of supporters began arriving at Safidhanj within days, and Abu Muslim ordered its fort repaired and strengthened.

Who actually turned out

The ethnic composition of those first adherents is debated. Scholarly accounts emphasise Arab settlers alongside Persian mawali rather than an overwhelmingly Persian force, and Abu Muslim's own origins remain uncertain.

Encyclopaedia Iranica identifies 2,200 early supporters as joining in the opening phase.

The consulted primary and scholarly sources do not substantiate a figure near 10,000 present when the banners first went up.

Walking into the provincial capital

One modern scholarly synthesis dates Abu Muslim's occupation of Merv to 14 or 15 February 748, and reports that he entered without opposition after waiting for fighting to begin between Nasr ibn Sayyar and Ali ibn Juday al-Kirmani.

Nasr, the last Umayyad governor of Khurasan, fled the city with a small escort and withdrew toward Nishapur.

The unresolved factional war between those two rivals was the opening; this was a political occupation, not a pitched battle in which Abu Muslim defeated Nasr.

A government by evening

Abu Muslim occupied the government palace, accepted allegiance and installed officials — a guard chief, a security chief, a judge and a registry official.

He also executed leading officers associated with the previous order.

A later narrative puts the number of those officers at twenty-four.

The date, and what came after

The exact date is genuinely disputed. Scholarly works give 14–15 February 748, 27 January 748, 12 or 14 January, and further variants across December 747 and January 748.

Those disagreements reflect variant lunar-month readings and conversions of dates preserved in medieval narratives, so 14 February should not be presented as uncontested.

Qahtaba's pursuit of Nasr, the taking of Nishapur, the battle at Jurjan and the occupation of Rayy were later stages of the 748 campaign; taking Merv consolidated the revolt in Khurasan without yet overthrowing the Umayyad caliphate.

A choice made at swordpoint in 749

Khurasani revolutionary leaders discovered the Abbasid family in hiding at Kufa in 749, armed themselves, went to its residence and swore allegiance to Abu al-Abbas Abd Allah ibn Muhammad — forcing the revolutionary administrator Abu Salama to accept their choice.

At a Friday gathering in Kufa's Great Mosque, Abu al-Abbas mounted the pulpit and presented the family's claim to rule through descent from al-Abbas, the Prophet Muhammad's paternal uncle.

He then became unable to continue — the accounts describe fever, illness or nervousness — and his uncle Dawud ibn Ali completed the address.

No day survives

The listed 28 October is not securely supported, and it fell on a Tuesday, whereas the narrative traditions place the mosque proclamation on a Friday.

Published reconstructions vary sharply: 3 Rabi I / 20 October, 13 Rabi I / 30 October, about 14 Rabi I / 31 October, and 12 Rabi II / 28 November.

October 749 is the safest month-level date; the variations may reflect different source traditions, manuscript readings, or the separation of an initial oath from a later public ceremony.

Two caliphs at once

The new caliph promised better times, and allegiance was received from the assembled Kufans and Khurasani soldiers. Surviving versions of the speech associate him with the title al-Saffah.

That title is contested. It can evoke pouring out blood, but scholars have also argued for the generous one — one who pours out gifts.

The ceremony did not end Umayyad rule: Marwan II remained caliph with substantial territory and armies, and the decisive battle at the Zab had not yet been fought.