Read history · Notable stories

AD 880

Battle of Hafrsfjord

The only early poem about the battle says the defeated ships came from the east — the opposite direction from the familiar story.

A date with no medieval basis

The traditional 18 July 872 does not hold. No medieval source supplies the day, and published estimates for the year range from about 868–875 through the 880s to around 900.

What the sources support is that late in the ninth century Harald defeated an alliance of petty kings and chieftains in a sea battle in Hafrsfjord, southwest of present-day Stavanger.

The early poem Haraldskvæði names two opponents by epithet, Kjotve and Haklang, and says Haklang fell in the fighting.

Which way the ships came

The poem describes ships arriving from the east to fight Harald. Recent source-based interpretation treats him as a southwestern ruler attacked by forces from farther east, rather than an eastern conqueror finishing a march across Norway.

The trigger is not securely preserved. A scholarly reconstruction proposes rivalry over coastal control and the trade corridor linking southwestern Norway with Kaupang and Hedeby.

Haraldskvæði may be near-contemporary, but it survives inside much later manuscripts, and its attribution — to Þorbjǫrn Hornklofi or to Þjóðólfr of Hvinir — is not entirely certain.

What the victory secured

The opposing force was driven off, and Harald's control over southwestern Norway and the Stavanger–North Jæren maritime corridor was strengthened. The sources do not support an immediate proclamation over all Norway.

The idea that this one battle united all of modern Norway was developed by twelfth-century historians roughly three centuries later, and no contemporary evidence records a coronation.

The story that oppressive taxation afterwards drove the settlement of Iceland is a later saga foundation narrative, now studied as myth.

Researched 23 Aug 2026 6 sources not yet audited Traditional date

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Traditional date. Conventionally dated here; the research places it at ~0880.

What the sources leave uncertain

No medieval source supplies 18 July; the day is unattested.
The year 872 is traditional. Published estimates range from approximately 868–875 through the 880s to about 900.
Haraldskvæði may be near-contemporary, but it survives inside much later manuscripts, and its authorship and interpretation are debated.
The scale of Harald’s realm and the battle’s role in Norway’s eventual political consolidation remain disputed.

Checked against

October 26, AD 899

Alfred the Great

Alfred's death date survives as a countdown to a feast: six nights before All Hallows.

26 October 899

Alfred died on 26 October 899. Tenth-century calendars record the day, and the Parker Chronicle describes it as six nights before All Hallows.

Medieval manuscripts label the annal 900 or 901, because of differing year starts and later annal numbering. Reign-length calculations and the Chronicle's September year-start support the modern date.

The Chronicle gives his reign as 28½ years.

What the sources leave out

The annalistic and calendar notices record the death but preserve no medical cause.

They establish his burial at Winchester, but not where he died.

He was not called the Great in his lifetime; recent scholarship traces the established sobriquet to the sixteenth century.

A contested succession

The Anglo-Saxon Chronicle says Edward, Alfred's elder son, then succeeded to the kingdom.

The succession was challenged almost at once by Alfred's nephew Æthelwold, son of his elder brother Æthelred.

Æthelwold seized the royal residences at Wimborne and Christchurch against Edward and his council, then fled by night.

Researched 24 Aug 2026 6 sources not yet audited

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

What the sources leave uncertain

The day, 26 October, is strongly supported by multiple calendars and the Chronicle’s relation to All Saints’ Day.
Medieval manuscripts label the annal 900 or 901 because of differing year starts and later annal numbering. Reign-length calculations and the Chronicle’s September year-start support the modern date 899.
The sources establish Alfred’s burial at Winchester but do not securely identify where he died or what caused his death.

Checked against

AD 900

Oleg of Novgorod

The founding story turns on an ambush: warriors hidden in boats, two rulers called out to greet supposed merchants, and a child produced as proof of legitimacy.

The chronicle's 882

The Primary Chronicle places the episode in 882: Oleg travelled down the Dnieper with the young Igor and a composite force, taking Smolensk and Liubech before reaching Kyiv.

In the narrative he hid warriors in the boats, presented himself as a traveller or merchant bound for the Greeks, lured Kyiv's rulers Askold and Dir outside, displayed Igor as Rurik's son, and had the two killed.

The Chronicle then says Oleg established himself as prince there and called Kyiv the mother of Rus' cities.

Two centuries of distance

No contemporary account dates the seizure to 882; the Primary Chronicle was compiled in the early twelfth century.

The Novgorod First Chronicle lacks the date entirely and differs over whether Oleg or Igor headed the expedition.

Modern scholarship accepts that a violent change of power probably occurred, but places the Rus' takeover of the middle Dnieper around the turn of the tenth century, with estimates ranging from the 890s into the 910s.

What the story is doing

The merchant disguise and the dynastic speech contain recognized folkloric and legitimizing motifs, even if they preserve memories of a real takeover. Kyiv already had rulers in the narrative, so Oleg was not its first.

The oldest cited dendrochronological date for buildings in Kyiv's Podil district is 887.

Reading the moment as the formation of Kievan Rus' compresses a long process of urbanization, tribute collection, trade and political consolidation into a single act.

Researched 24 Aug 2026 5 sources not yet audited Date disputed

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Date disputed. Sources disagree; the research places it at ~0900.

What the sources leave uncertain

No contemporary account dates the seizure to 882. The Primary Chronicle was compiled in the early twelfth century.
The Novgorod First Chronicle lacks the 882 date and differs over whether Oleg or Igor headed the expedition.
A modern scholarly reconstruction places the Rus’ takeover of the middle Dnieper around the turn of the tenth century; estimates range from the 890s into the 910s.
The story’s merchant disguise and dynastic speech contain recognized folkloric and legitimizing motifs, even if they may preserve memories of a real violent takeover.

Checked against

AD 900

Classic Maya collapse

Around the supposed endpoint of Classic Maya civilization, Uxmal's hieroglyphic writing was peaking and Chichén Itzá's biggest ceremonial complex was still going up.

900 is a line in a textbook

No identifiable Maya-wide event occurred in 900. Roughly 900 or 1000 is a modern, conventional boundary between archaeological periods, and scholars debate both its timing and its usefulness.

One major synthesis uses a Terminal Classic span of 800 to 1000 CE. Others end the Classic around 900, around 1000, or run the Terminal Classic into the eleventh century.

Site chronologies are continually revised through radiocarbon dating, dendrochronology, inscriptions, ceramics and paleoclimate records.

What changed, and where

Across much of the southern and central lowlands, the eighth through tenth centuries brought the dissolution of divine kingship, declining production of royal inscriptions and monumental architecture, political fragmentation, population movement and abandonment — at different times in different places.

Tikal, Palenque, Copán, Uxmal and Chichén Itzá did not decline together. Uxmal's hieroglyphic activity is estimated to peak around 895–907, with the end of its political system centred about 950.

Chichén Itzá became the dominant northern-lowland centre during the Terminal Classic, and much of its northern ceremonial complex — El Castillo included — was built after 900.

Drought, and the limits of the word collapse

One Terminal Classic paleoclimate record resolves eight severe droughts lasting three to eighteen years, with precipitation 36 to 52 percent below long-term averages. The longest in the recent Uxmal-area reconstruction ran thirteen years, from 929 to 942.

Research identifies interacting, regionally variable pressures: recurrent drought and changing rainfall predictability, environmental and agricultural stress, warfare, altered trade, demographic pressure, weakening elite authority and decisions to migrate. No single causal theory accounts for the timing and severity everywhere.

The Maya did not disappear. Collapse chiefly describes political and demographic transformations in particular lowland regions, especially the fall of divine kingship.

Researched 24 Aug 2026 6 sources not yet audited Date disputed

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Date disputed. Sources disagree; the research places it at ~0900.

What the sources leave uncertain

The year 900 is approximate periodization, not a dated collapse event.
Scholars variously end the Classic period around 900, 1000, or use a Terminal Classic extending into the eleventh century.
Site chronologies are continually revised through radiocarbon dating, dendrochronology, inscriptions, ceramics, and paleoclimate records.
No single causal theory accounts for the timing and severity of change across every Maya region.

Checked against

January 5, AD 910

Abdallah al-Mahdi Billah

A single Friday sermon changed the ruler named in public prayer across Qayrawan and Raqqada.

Three ceremonies, not one

The winter usually filed under 909 spreads across several occasions. Abd Allah al-Mahdi was acclaimed at Sijilmasa on 27 August 909, then entered the former Aghlabid palace city of Raqqada on 4 January 910 and was acclaimed caliph there.

At the first Friday sermon after his arrival, on 5 January, congregations in Qayrawan and Raqqada heard him proclaimed Amir al-Mu'minin and Imam al-Muslimin — Commander of the Faithful and Imam of the Muslims — with the regnal title al-Mahdi bi'llah.

Some translations and Hijri conversions give 7 January for the Raqqada proclamation; January 910 is secure, and the precise day depends on which stage and textual tradition is meant.

The state had already been won

Abu Abd Allah al-Shi'i and the Kutama Berbers had overthrown the Aghlabids in March 909, then brought the previously concealed imam from Sijilmasa so that the deputy government could be placed under his direct authority.

His arrival therefore transferred authority rather than beginning a conquest: Fatimid rule in Ifriqiya passed from a missionary-led provisional regime to the publicly declared imam-caliph.

Abu Abd Allah had governed as al-Mahdi's deputy, and escorted him from Sijilmasa.

What the sermon said

The sermon publicly asserted that legitimate rule and the imamate had returned to Muhammad's family, formally inaugurating Fatimid government in North Africa.

It invoked Fatima, Ali, Hasan and Husayn, making the new regime's Shi'a claim audible to the assembled congregations.

Later purges, revolts, Egyptian expeditions and the construction of Mahdiya belong to other dates entirely.

Researched 23 Aug 2026 3 sources not yet audited Date disputed

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Date disputed. Sources disagree; the research places it at 0910-01-05.

What the sources leave uncertain

The page's listed year, 909, conflates several ceremonies. Al-Mahdi was acclaimed at Sijilmasa on 27 August 909, entered Raqqada on 4 January 910, and was named with his full titles in the Friday sermon on 5 January.
Some translations or Hijri-calendar conversions give 7 January for the Raqqada proclamation. January 910 is secure; the precise day depends partly on which stage and textual tradition is meant.

Checked against

AD 911

Treaty of Saint-Clair-sur-Epte

The best near-contemporary proof is half a sentence buried in a royal charter issued about seven years later.

What can be recovered

Around 911, Charles III granted Rollo and his followers Frankish territory centred on Rouen. That is the recoverable core of the event traditionally called the Treaty of Saint-Clair-sur-Epte.

No treaty text or contemporary narrative of the negotiation survives, so the familiar label can wrongly suggest historians possess one.

The earliest surviving documentary confirmation is Charles's charter of 14 March 918, which excepted land granted to the Seine Normans — specifically Rollo and his companions — for the protection of the kingdom.

What the charter does not say

It confirms an earlier grant but states no date, no boundaries, no baptismal conditions, no homage ritual, and no title for Rollo.

Homage, baptism, the meeting at Saint-Clair-sur-Epte and the detailed territorial terms come principally from Dudo of Saint-Quentin, writing roughly a century later for the Norman ruling house.

The year, the place and even the season rest chiefly on that later account.

Not a duchy yet

Rollo was not demonstrably made first Duke of Normandy. His contemporary title is unknown, and Count of Rouen is undocumented for him.

The Epte-to-Risle boundary, the division among chieftains, the rejected grant of Flanders and the personal homage ceremony all depend on later narrative tradition.

Conversion was uneven as well: a papal response from the 910s discusses baptized Northmen who continued pagan practices.

Researched 23 Aug 2026 5 sources not yet audited Traditional date

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Traditional date. Conventionally dated here; the research places it at ~0911.

What the sources leave uncertain

No treaty text or contemporary narrative of the negotiation survives.
The year 911 and location Saint-Clair-sur-Epte depend chiefly on Dudo's later account. Even 'autumn' is a reconstruction, not a documented date.
The 918 charter confirms an earlier grant to Rollo and his companions but does not state its date, boundaries, baptismal conditions, homage ritual, or Rollo's title.

Checked against

July 25, AD 918

Taejo of Goryeo

The chronicle says Wang refused the throne until his wife carried out his armour and put it on him herself.

The night before

On the night before 25 July 918, the cavalry commanders Hong Yu, Pae Hyeon-gyeong, Sin Sung-gyeom and Pok Ji-gyeom secretly approached Wang Geon's house and urged him to accept the throne.

The Goryeosa says Wang initially refused, but his wife Lady Yu brought his armour and put it on him; the commanders then escorted him out as supporters gathered at the palace.

Messengers rode through Cheorwon announcing that Lord Wang had raised a righteous banner, and a reported crowd of more than 10,000 gathered at the palace gates.

The ruler who ran

The conspiracy had grown out of opposition to Kung Ye's violent purges and increasingly coercive rule.

Kung Ye fled Cheorwon through the north gate in disguise, hid for two nights, and was killed by people at Buyang.

The capital stayed at Cheorwon through the accession; Wang moved it to Songak, near modern Kaesong, in 919.

Enthroned in Pojong Hall

On the fifteenth day of the sixth lunar month — converted as 25 July 918 — Wang was enthroned in Pojong Hall, named the state Goryeo and adopted the era name Cheonsu.

The following day he issued an accession edict condemning Kung Ye's cruelty, burdensome labour and taxes, and the killing of his own family, while promising general renewal.

Two days after taking the throne he ordered the release of more than 80 soldiers whom Kung Ye had summoned in chains despite their innocence.

Researched 23 Aug 2026 5 sources not yet audited

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

What the sources leave uncertain

The exact solar date is a modern conversion of the Goryeosa's lunisolar entry: sixth month, fifteenth day. The National Institute of Korean History gives 25 July 918.
The detailed coup narrative comes from the Goryeosa, compiled in the fifteenth century from earlier records; colorful details such as Lady Yu producing the armor should therefore be attributed to that chronicle.
The conspiracy began on the preceding night; 25 July is specifically the date of enthronement and naming of Goryeo.

Checked against

January 16, AD 929

Caliphate of Córdoba

A caliphate was announced through bureaucratic mechanics: letter headings, and the name spoken from every Friday pulpit.

A circular dated Thursday

Abd al-Rahman III issued a proclamation dated Thursday, 2 Dhu'l-Hijja 316 — conventionally converted as 16 January 929 — ordering that he be styled Commander of the Faithful.

He directed officials to use the new title in correspondence and instructed local preachers to pronounce it in the Friday khutba.

The text asserted that anyone else using the title did so unlawfully, and that continuing not to claim it would mean surrendering an Umayyad right.

Rank, not independence

This was a change in claimed religious-political rank rather than a declaration of political independence from Baghdad; the Umayyad emirate in al-Andalus had been independent for generations.

Nor was it a break with the Fatimids, who were rivals Córdoba had never owed allegiance to. Scholars set the decision against Fatimid expansion in North Africa and Abbasid weakness.

It was transmitted as an administrative circular directing titulature and public prayer, not securely as a dramatic personal speech in the Great Mosque.

The second badge of sovereignty

By taking control of the khutba, and later that year of coinage in his own caliphal name, Abd al-Rahman formally raised the polity from emirate to caliphate.

He adopted the caliphal style Amir al-Mu'minin alongside the regnal name al-Nasir li-Din Allah.

The Córdoba mint's gold dinars belong later in 929, not to 16 January.

Researched 23 Aug 2026 4 sources not yet audited

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

What the sources leave uncertain

The proclamation itself records the Hijri date 2 Dhu'l-Hijja 316; 16 January 929 is its conventional Common Era conversion.
The exact sequence between the dated circular and the first Córdoba khutba is rendered differently in later chronicles, but the date, new title, and instructions are well supported.

Checked against

January 11, AD 930

Qarmatian sack of Mecca

An army entered Islam's holiest city posing as pilgrims, then attacked the people circling the Kaaba.

7 Dhu al-Hijja 317

On 7 Dhu al-Hijja 317 AH — equated by the Encyclopaedia of Islam entry to 11 January 930 — Abu Tahir's Qarmatian force reached Mecca posing as pilgrims and gained access to the sanctuary.

The local attempt to keep it out was overcome by asserting a Muslim's right of entry. Once inside, the Qarmatians massacred pilgrims and inhabitants and plundered Mecca and the Kaaba.

The pillaging continued for about a week.

Zamzam, and the door of the Kaaba

Accounts report that victims' bodies were thrown into the Zamzam Well, from which pilgrims drew sacred water.

The Kaaba's door and its covering were removed.

No reliable casualty total survives; the figures of 20,000 or 30,000 that circulate in popular retellings have no dependable basis.

The stone goes east

The attackers dislodged the Black Stone and carried it to their capital, al-Ahsa, in eastern Arabia. They did not carry away the Kaaba itself.

The Qarmatians of Bahrayn were an eastern-Arabian polity: medieval Bahrayn meant a larger mainland region including al-Ahsa and Qatif, not today's island state.

The Stone's restoration is dated 339 AH, rendered 950–951 — an absence of about twenty-two lunar years.

Researched 23 Aug 2026 5 sources not yet audited

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

What the sources leave uncertain

The date 7 Dhu al-Hijja 317/11 January 930 is supported by the Encyclopaedia of Islam entry, although medieval lunar-date conversions can vary by a day between modern systems.
Detailed speeches, poems and dramatic incidents survive through later narrative sources and should be introduced as reports, not eyewitness certainties.
No reliable casualty total was found; figures such as 20,000 or 30,000 in popular retellings should not be used.

Checked against

AD 939

Ngô Quyền

After winning a river battle, his recorded first acts were strikingly administrative: name a queen, appoint officials, colour-code the official robes.

Spring 939

In spring 939, following his victory over Southern Han, Ngô Quyền began calling himself king.

The Vietnamese chronicle Đại Việt sử ký toàn thư records that he made Lady Dương queen, appointed officials, established court rituals and designated the colours of official robes.

The chronicle gives only a season, and says he called himself king — it does not describe a coronation on a known day.

Cổ Loa, not Đại La

He chose Cổ Loa — the former capital associated with the ancient Âu Lạc polity — as his capital and court seat, instead of the former Chinese administrative centre at Đại La.

The choice carried symbolic weight, though that should not be inflated into a documented revival of prehistoric Lạc Việt institutions.

With a king, a queen, appointed officials, prescribed ritual and robe colours, the territory was no longer being presented as a Chinese-style military governorship.

What the victory did and did not end

The destruction of the Southern Han invasion fleet at Bạch Đằng in late 938 removed the immediate external challenger and allowed him to replace the inherited military-governor framework with openly royal government.

It consolidated autonomy and prevented Southern Han reconquest. Calling it the end of a thousand uninterrupted years of Chinese rule obscures earlier revolts, autonomous governments and the collapse of direct Tang administration before 938.

His reign ran from 939 to 944.

Researched 23 Aug 2026 5 sources not yet audited

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

What the sources leave uncertain

The sources support spring 939 but do not provide a secure month and day; precise dates sometimes found online are not established by the cited chronicle.
The principal narrative chronicle was compiled centuries later, so its institutional details are evidence for the received Vietnamese historical tradition rather than a contemporary administrative record.
Ngô Quyền's birth year is variously rendered 897 or 898 in modern Vietnamese sources.

Checked against

AD 941

Occultation (Islam)

The reported final message named no successor. It told the intermediary that he would be dead in six days and must appoint no one.

The year is 941, not 939

The date 939 appears in some derivative references but conflicts with the standard chronology. Twelver tradition places the end of the Minor Occultation in 329 AH, a year overlapping 940 and 941, which scholarship prints as 329/940–41 or simply 941.

In that tradition, Ali ibn Muhammad al-Samarri, the last recognized intermediary of the Hidden Imam, produced a decree in 329 AH saying he would die in six days and must appoint no successor.

Al-Samarri died that year, and the succession of official intermediaries ended.

An office that simply stopped

The cessation of regular communication through an authorized intermediary became the boundary between the Minor and Major Occultations in Twelver doctrine.

This was not a newly observed physical disappearance: it was the end of the claimed intermediary system for an Imam already believed to be hidden.

The precise formulation is that no regular authorized intermediary remained — later Twelver devotion includes private encounter traditions without recognizing a new official deputy.

How the chain was assembled

Twelver tradition retrospectively recognizes four authorized mediators — Uthman ibn Sa'id, Muhammad ibn Uthman, Husayn ibn Ruh and Ali al-Samarri — spanning 260 to 329 AH, about seventy years.

That division into Minor and Major Occultations, and the canonical chain of four, was systematized retrospectively.

Encyclopaedia Iranica notes that whether Hasan al-Askari left a son remains disputed, and that traditional accounts of the four deputies' activities are partly historically unsound.

Researched 23 Aug 2026 6 sources not yet audited Traditional date

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Traditional date. Conventionally dated here; the research places it at 0941.

What the sources leave uncertain

The event is principally a claim of Twelver religious tradition. Encyclopaedia Iranica notes that whether Hasan al-Askari left a son remains disputed and that traditional accounts of the four deputies' activities are partly historically unsound.
The Islamic year 329 AH overlaps 940–941 CE and scholarship consequently prints 329/940–41 or simply 941. The supplied date 939 is not supported.
The exact reported decree and its six-day prediction survive in later Twelver sources; they should be attributed rather than narrated as independently verified correspondence.
The standard division into a Minor and Major Occultation and the canonical chain of four deputies became systematized retrospectively.

Checked against

AD 946

Millennium Eruption

A temple chronicle in Nara, about a thousand kilometres away, described white ash falling like snow through the night.

Late 946, and a day inferred

Tree-ring radiocarbon evidence securely places the Millennium Eruption in late 946. The 2 November date is an inference: a Japanese report of ash on 3 November suggests a principal eruptive phase may have begun about a day earlier.

On the night of 3 November 946, a temple chronicle in Nara recorded white ash falling like snow. Nara lies roughly 1,000 kilometres from Paektu.

The dating paper explicitly cautions that the historical report is not certain proof of Paektu ash.

How the year was fixed

A subfossil larch killed and buried by the eruption preserved the 774 CE cosmogenic radiocarbon spike, which allowed its final growth ring to be fixed to 946.

Paektu produced Plinian fallout and pyroclastic-density currents during a large, compositionally zoned explosive eruption that helped form the present summit caldera.

Modeling puts each of the two principal fallout phases at an eruption column roughly 30 to 40 kilometres high, lasting about fifteen and twenty hours.

One date cannot hold it

Recent ice-core work indicates two main eruptive phases separated by roughly one or two months, so a single date cannot stand for the whole eruption.

The explosivity is unsettled. The Smithsonian database currently assigns VEI 6 and cites a reassessment rejecting VEI 7, while some recent papers retain a 6-or-7 range; estimated erupted magma volume is about 23 cubic kilometres, with a modelled range of 13 to 47.

Size did not translate into severe cooling: Greenland ice and Northern Hemisphere tree rings show weak sulfur deposition and no statistically significant summer cooling signal.

Researched 23 Aug 2026 5 sources not yet audited Date disputed

How this was checked

Researched from the web into a fact sheet and rewritten from that sheet. This article has not yet had an independent model audit. How the pipeline works →

Date disputed. Sources disagree; the research places it at ~0946-11-02.

What the sources leave uncertain

Late 946 is secure, but 2 November is an inference from ash reported at Nara on 3 November and an estimated transport time of less than a day. The dating paper explicitly cautions that the historical report is not certain proof of Paektu ash.
Recent ice-core work indicates two main eruptive phases separated by roughly one or two months. Consequently, a single date cannot represent the entire Millennium Eruption.
The event's explosivity is still expressed as VEI 6 or 7 in recent literature. The Smithsonian database currently assigns VEI 6 and cites a dedicated reassessment rejecting VEI 7.
Reports of thunder-like sounds in Korea in 946 and Japan in February 947 are compatible with eruption phenomena but cannot be conclusively attributed to Paektu.