Battle of Talas 751: Tang–Abbasid War, Karluks and the Paper Myth
A source-critical history of the Battle of Talas in 751: Gao Xianzhi, Ziyad ibn Salih, Karluk participation, uncertain armies, Tang strategy, Abbasid Central Asia and why one battle did not simply carry papermaking west.
Battle of Talas: facts at a glance
| Question | Evidence-based answer |
|---|---|
| When? | July 751 CE; exact day is not securely established. |
| Where? | In the Talas River–Taraz landscape, probably near Atlakh; the precise battlefield remains debated. |
| Who fought? | A Tang army under Gao Xianzhi against an Abbasid-aligned force under Ziyad ibn Salih, with Karluks and other Central Asian actors. |
| Who won? | The Abbasid-aligned coalition; Gao escaped with a remnant of his force. |
| Why did it happen? | A local contest involving Shash, Ferghana and Tang expansion intersected with new Abbasid power in Khurasan and Transoxiana. |
| Did it end Tang power? | No. Tang influence and diplomacy continued; the An Lushan rebellion from 755 and several frontier wars mattered more to withdrawal. |
| Did it spread Islam at once? | No. Islamization among Turkic communities was uneven and gradual. |
| Did captives introduce paper? | They may have transferred skills, but paper and probably production existed in Central Asia earlier; the neat origin story is late. |
What happened at Talas in 751?
A Tang expedition operating beyond the Pamirs met an army linked to the recently established Abbasid Caliphate in the Talas valley. Gao Xianzhi’s force had advanced through a political landscape of oasis kingdoms, mountain routes and mobile confederations. Ziyad ibn Salih’s side drew on the military resources of Abbasid Khurasan and on regional partners. During the fighting, Karluk forces are reported to have turned against or abandoned the Tang side. Gao’s army collapsed, and he withdrew with only part of it.
That concise account contains nearly everything that is reasonably secure. The precise site, the order of battle, the timing of the Karluk move, the number of combatants and even some command relationships are reconstructed from later and uneven sources. Talas was therefore a real Abbasid victory and a serious Tang defeat, but the confidence of a modern tactical map often exceeds the surviving evidence.
Where was the battlefield? Talas, Taraz and Atlakh
The battle is named for the Talas River system in what is now the border region of Kazakhstan and Kyrgyzstan. Medieval Taraz, written in several forms in Chinese, Arabic and Persian sources, was an important town on routes connecting the Syr Darya basin, the Chuy valley, Ferghana and lands east of the Tian Shan. Atlakh, a settlement or district near Taraz, is frequently proposed as the immediate battlefield.
Archaeology and landscape research can narrow plausible corridors, but they have not produced a universally accepted line of battle. Rivers shifted, settlements changed and texts were not written as modern grid references. It is sound to say ‘the Talas–Taraz region, probably near Atlakh’ and unsound to present an exact tourist marker as proven. This geographical caution does not reduce the event’s importance; it separates knowledge from commemoration.
Central Asia before the battle was not empty space
Maps often color Tang territory in one block and Abbasid territory in another, making the two empires appear to share a hard modern border. Eighth-century Central Asia worked differently. Sogdian cities, Turkic confederations, Tibetan power, the kingdoms of Ferghana and Shash, and Chinese garrisons negotiated, fought, traded and changed allegiance. Imperial titles could coexist with local rulers and tribute relationships; control weakened sharply with distance and season.
This setting explains why Karluks, rulers of Shash and Ferghana, and other groups are not secondary extras. They controlled pasture, passes, supplies, intelligence and cavalry networks. Gao and Ziyad entered a regional contest whose participants pursued their own survival and advantage. Calling Talas simply ‘China versus the Arabs’ erases the people on whose land the campaign occurred and mistakes imperial labels for homogeneous armies.
The road to Talas: Ferghana, Shash and Gao Xianzhi
Gao Xianzhi was a Goguryeo-descended Tang general who had built his reputation through difficult campaigns in the high routes west of the Tarim Basin. Tang sources credit him with bold operational movements, but his intervention in local affairs also generated enemies. A dispute between Ferghana and Shash became the immediate path to Talas. Gao supported Ferghana, attacked Shash and captured its ruler after an arrangement that later accounts describe as violated.
A surviving prince of Shash sought support from powers to the west. The story is preserved with moral shaping, and its details should not be treated as a verbatim diplomatic transcript. Even so, it shows that Talas did not begin because an Abbasid caliph and a Tang emperor planned a world-historical showdown. It emerged from a chain of local appeals, punitive campaigns, contested promises and imperial opportunity.
Why the Abbasids were active beyond the Oxus
The Abbasids had overthrown the Umayyads only in 750. Their revolution was rooted heavily in Khurasan, where Arab settlers, Iranian populations, military clients and diverse opponents of Umayyad rule formed an unstable coalition. Establishing authority in Marw, Bukhara, Samarkand and other eastern centers required commanders to bargain with local elites while suppressing rebellions and rival claimants.
Ziyad ibn Salih operated within this unsettled consolidation. The Abbasid presence at Talas was neither a fully mature national army nor an accidental raiding party. It represented a new regime testing its eastern reach through Khurasani forces and alliances. The victory gave that regime prestige and removed an immediate Tang expedition, yet Abbasid rule itself remained contested. Ziyad later rebelled and was killed, a reminder that the victors were not politically unified for long.
The armies: what can and cannot be counted
Traditional figures range from tens of thousands to far larger totals, while reports of prisoners and dead can be equally dramatic. Such numbers served literary purposes: magnifying victory, explaining disaster or demonstrating the scale of imperial power. They also mix regular soldiers, allied cavalry, local levies and support personnel. No pay register or complete muster survives that permits a precise modern total.
A cautious reconstruction recognizes a substantial Tang-led expedition with Chinese regulars and allied contingents facing a substantial Abbasid-aligned regional army. The Karluks may initially have been associated with Gao, may have maintained an ambiguous position, or may have coordinated with his opponent before the battle; sources do not allow a clean answer. The balance of local cavalry and the vulnerability of Gao’s communications probably mattered more than a suspiciously exact five-digit figure.
How the battle may have unfolded
Later narratives describe several days of combat before the Karluks attacked the Tang force from the rear while the Abbasid-aligned army pressed from the front. The resulting encirclement or collapse inflicted heavy losses. Gao escaped with the help of officers including Li Siye, whose actions are celebrated in Chinese biographical traditions. This broad sequence is plausible and widely repeated.
It should not be embellished into a minute-by-minute certainty. We cannot securely map every cavalry charge, formation or speech. The language of ‘betrayal’ also assumes a stable contract that may not have existed. Karluk leaders made a political choice in a fluid environment, and their later history cannot be reduced to that moment. The best explanation combines battlefield pressure, shifting allegiance and Gao’s long, exposed route rather than a single surprise trick.
Why the Tang did not return with another great expedition
Talas destroyed Gao’s immediate campaign, but it did not make Tang power vanish west of the Pamirs overnight. The court still received embassies, maintained relations with Central Asian rulers and possessed garrisons in the Tarim Basin. Four years later, however, the An Lushan rebellion erupted. It consumed enormous resources, devastated the imperial heartland and forced the redeployment of frontier troops.
At the same time, the Tibetan Empire pressed into strategic corridors, Nanzhao challenged Tang power in the southwest, and the dynasty’s fiscal and military structure changed. Central Asian partners adapted to the new balance. Talas was one warning and one lost expedition inside this larger contraction. If a single date must explain the Tang retreat, 755 is at least as important as 751—but the better account uses a sequence, not a substitute monocause.
What the Abbasid victory changed immediately
The most direct result was operational: Gao’s army was broken, the threat it posed to Shash and neighboring polities receded, and Abbasid-aligned forces gained prestige. The outcome demonstrated that Tang expeditions could be defeated far from their strongest supply bases. It also gave later writers a convenient milestone for the transition from Tang to Islamic political predominance in Transoxiana.
Yet the victors did not annex a neat block of territory after the battle, and the caliphate did not create a fixed frontier at Talas. Local dynasts remained important, revolts continued and Abbasid governors had limited capacity. The event helped a trend, but it was not equivalent to modern conquest followed by cadastral administration. Immediate military advantage and long-term cultural transformation must be kept separate.
Did Talas cause the Islamization of Central Asia?
No mass conversion followed the battlefield. Muslim communities and rulers were already present in parts of Central Asia, while Zoroastrian, Buddhist, Christian, Manichaean and local religious traditions remained active. Islam expanded through urban communities, taxation and legal incentives, patronage, intermarriage, missionary activity, trade and the choices of later Turkic rulers. The process differed by town, oasis and confederation.
Major Turkic conversions associated with the Qarakhanids belong to a much later context. Even political submission to a Muslim governor did not automatically mean religious conversion. Talas may have strengthened conditions favorable to Islamic institutions in some western cities, but treating defeated soldiers as an entire civilization changing faith confuses military and social history.
The famous story of Chinese papermaker captives
A popular narrative says that Chinese papermakers captured at Talas revealed a guarded secret, founded a mill at Samarkand and thereby transmitted paper to the Islamic world and Europe. The story is memorable because it gives a global technology one battle, one group of prisoners and one direction of travel. Its evidentiary foundation is weak. The explicit captive tradition appears centuries after the battle, and specialist histories repeatedly warn that it has been treated as established fact without contemporary support.
Captives certainly moved after medieval wars, and craftspeople could carry practical knowledge. It is therefore possible that people taken in 751 contributed skills or improved local practice. What is not justified is the claim that nobody west of China knew paper or papermaking beforehand. Fragments, documents and trade indicate earlier circulation, and Central Asian production may predate the battle. Possible contribution is not the same as unique invention.
Paper existed in Central Asia before 751
Paper traveled west from its early development in China along networks that also carried textiles, medicines, metals, religious texts and administrative habits. Archaeological finds from Central Asian sites contain paper older than the battle. Some sheets were imports; others may indicate local work. Because laboratories cannot always identify a workshop, evidence for use is stronger than evidence for the exact date of manufacture, but it decisively breaks the claim that Talas introduced the material itself.
Samarkand’s location made it a natural production and distribution center whether or not one royal workshop opened in 751. Water, urban demand, textile waste and merchant routes supported the craft. The historically interesting question is therefore not ‘who stole the secret on one day?’ but how scattered knowledge became sustained manufacture, how techniques changed with available fibers and how buyers created an expanding market.
How Islamic-world papermaking adapted the technology
Chinese papermakers commonly used bast fibers and other plant material. In Central and western Asia, producers increasingly exploited linen and hemp rags, pounding and preparing fibers according to local resources. Sizing, polishing, sheet formats and quality grades also developed in response to scribal requirements. These were not passive copies: they were technical and commercial adaptations.
By the late eighth century paper manufacture is associated with Baghdad, and paper use expanded across administration, scholarship and trade. Production later appears in Egypt and Islamic Iberia. The sequence spans generations and many workshops. Paper did not automatically create an intellectual ‘golden age,’ but lower cost and greater availability than parchment or papyrus could support bureaucratic records, bookselling and copying when combined with literacy, patronage and urban institutions.
Did Talas transfer other technologies?
Some retellings attach silk weaving, compass knowledge, gunpowder or printing to the same prisoners. This accumulation is a warning sign. Technologies rarely move as sealed packages, and there is no contemporary inventory of captured specialists. Silk production had already crossed borders through earlier migration and exchange. Printing developed through separate East Asian histories and reached western Asia under different conditions.
The responsible formulation is narrower: warfare could accelerate human movement, and the Talas campaign may have contributed to exchanges already underway. Each claimed technology needs its own dated evidence. A battle should not become a narrative container into which every later connection between East Asia and the Islamic world is placed.
Trade and diplomacy continued after Talas
Talas did not create an iron curtain across Eurasia. Merchants, envoys, pilgrims and displaced communities continued to travel. Tang China and the Abbasid world were connected indirectly and sometimes directly through maritime and overland routes. Goods could pass through many hands without the producer and final buyer sharing a border or political alliance.
The Sogdian commercial diaspora changed under conquest and rebellion but did not vanish in one year. Chinese textiles and ceramics reached western markets; glass, aromatics, metals and ideas moved east. Diplomatic missions continued because courts valued information and commerce even when their commanders had fought. The longer history after Talas is therefore one of altered connections, not separation.
Why Talas became larger in modern memory
For many medieval writers Talas was not the single hinge of world history it later became. Modern national histories, civilizational narratives and textbooks elevated it because it sits at an attractive intersection: Tang China, the early Abbasids, Turkic peoples and paper. A compact story can explain political change and technological transmission at once, which makes it easy to teach and easy to overstate.
Commemoration also reflects present geography. Kazakhstan and Kyrgyzstan claim the Talas landscape as part of national heritage; Chinese, Arab, Persian, Turkish, Japanese and European scholarship asks different questions of it. These perspectives can enrich the subject if their categories are not projected uncritically backward. The battle matters without needing to be the day ‘East met West.’ Central Asia had linked societies for centuries.
How historians reconstruct an event with uneven sources
Chinese dynastic histories and biographies preserve Gao’s campaign, defeat and officers’ conduct. Arabic and Persian traditions place the encounter within conquest and Abbasid eastern affairs. Both bodies of evidence were compiled with political and literary aims, often after the event. Place names shift across languages, and writers knew different sides of the campaign.
Historians compare chronology, geography, administrative context, archaeology and the later careers of participants. Agreement across independent traditions supports the core event, while contradictions require modest wording. Exact speeches, enormous numbers and clean causal chains receive less weight. Source criticism does not mean that nothing can be known; it allows secure claims to be distinguished from plausible reconstruction and attractive legend.
A balanced verdict on Talas
Talas deserves attention as a rare direct clash between Tang and Abbasid-aligned forces and as a decisive defeat for a major Tang expedition. It affected regional confidence, alliances and Gao’s western strategy. It also offers a valuable case study in how Central Asian actors shaped imperial outcomes.
Its global significance is cumulative rather than magical. Tang withdrawal resulted from the rebellion of 755 and multiple frontiers; Islamic political and religious predominance grew gradually; papermaking spread through pre-existing circulation, possible captive knowledge, workshop adaptation and demand. A smaller claim is not a weaker history. It is a richer account because it restores time, local agency and evidence.
Frequently asked questions
Who won the Battle of Talas?
The Abbasid-aligned coalition defeated Gao Xianzhi’s Tang expedition. This was a clear battlefield result, although it did not amount to the destruction of all Tang power in Central Asia.
Did the Karluks betray the Tang?
Several traditions make a Karluk attack or change of side decisive. ‘Betrayal’ is a moral label that assumes a stable alliance; their prior commitment and timing remain debated in a fluid regional system.
How many soldiers fought?
No exact total is secure. Large figures in later texts may include allies and support personnel and often serve rhetorical aims. Relative scale and coalition structure are more defensible than a precise number.
Did the battle bring papermaking to the Islamic world?
Not by itself. Paper circulated and probably was produced in Central Asia before 751. Captured craftspeople may have contributed knowledge, but the sole-origin story is late and unsupported by contemporary evidence.
Did Talas end Chinese rule in Central Asia?
No. Tang relations and positions continued after 751. The An Lushan rebellion from 755, Tibetan expansion and other military and fiscal pressures drove the larger retreat.
Why is the battle still important?
It was an unusual imperial encounter, a real Abbasid-aligned victory and a window into local Central Asian agency. Its historiography also teaches how one-battle explanations can distort long technological and religious changes.
Continue the history
- Islamic Golden Age timeline — paper, institutions and scholarship after the early Abbasids
- Islamic history timeline — place 751 within a wider chronology
- The Abbasid Revolution — how the new dynasty came to power in Khurasan
- Founding Baghdad in 762 — the capital where paper production later expanded
- The House of Wisdom — translation networks without the one-academy myth
- Al-Biruni — a later Central Asian scholar and cross-cultural method
Sources
This article prioritizes specialist and institutional research. The sources agree on the core defeat but differ on location, forces and causation; the papermaking claim is evaluated against material and documentary history rather than repeated as a battlefield anecdote.
- UNESCO Silk Roads: The Western Regions under the Tang Empire and Tibet — Used for Tang administration and rivalry in Central Asia and for the warning that Talas alone did not cause the later Tang retreat.
- Encyclopaedia Iranica: Paper and Papermaking — The principal source-critical reference for paper before 751, the late captive story, materials and the spread of production.
- Encyclopaedia Iranica: Paper in the Iranian World Prior to Printing — Used for documentary evidence, regional paper manufacture and the difference between the arrival of paper and local industrial expansion.
- Encyclopaedia Iranica: Khorasan in the Abbasid Period — Used for early Abbasid authority in Khurasan, local revolts and the unstable political setting east of the caliphate.
- Cambridge University Press: Tang China and the Rise of the Silk Roads — Used to place Tang expansion within Indian Ocean and overland networks rather than a closed China-versus-Islam frame.
- Qazaq Historical Review: The Historical Landscape of the Battle of Talas — Used for the Talas–Taraz landscape, competing site identifications and the limits of locating a precise battlefield.
- King’s College London: The Battle of Talas — A university overview used to cross-check commanders, Karluk participation and the battle’s later global reputation.
- CiNii Research: The Battle of Talas and Central Asia — Japanese academic bibliography used to check East Asian scholarship and Japanese terminology.
- OpenStax World History: East–West Interactions in the Early Middle Ages — Used as an open educational cross-check for the broad chronology, followed by correction from specialist paper scholarship.
- UNESCO: History of Civilizations of Central Asia, Volume IV — Used for the longer Central Asian transition, Abbasid governance, trade and urban development after the mid-eighth century.
