Kublai Khan sent an invasion fleet from Korea towards Japan in 1274. It reached Kyushu and fought Japanese defenders along the coast before withdrawing. In 1281 he launched a much larger expedition, which also failed. Storms at sea came to dominate accounts of both defeats, and Japanese religious traditions remembered the destructive winds as kamikaze, or divine winds. Yet the two campaigns did not unfold in the same way, and the weather was only part of what prevented the Mongol-led Yuan dynasty from conquering Japan.
Later retellings often credit both victories to typhoons, leaving the Japanese defenders with little part in the outcome. The 1281 catastrophe certainly involved a destructive storm, and underwater archaeology has since found physical remains of that disaster. The outcome of the first invasion is more disputed. Japanese forces had already fought the invaders before their departure in 1274, and the contemporary evidence does not securely establish the often-repeated story of an overwhelming typhoon destroying their fleet.
The campaigns involved fighting on small islands and along the Kyushu coast, supplies carried across the sea, Japanese walls and attacks on ships, crowded anchorages and severe weather. Accounts written on both sides can be compared with a samurai’s illustrated testimony, the excavated wrecks near Takashima and thirteenth-century flooding traces preserved in coastal sediments. Each kind of evidence can answer some questions better than others.
Why Kublai Khan Sent Fleets to Japan
Kublai Khan was Genghis Khan’s grandson. He controlled the Mongol empire’s eastern territories and declared the Yuan dynasty in China in 1271, ruling through Mongol elites and the populations and governments drawn into his power, including the Korean kingdom of Goryeo. Japan remained outside this imperial system. Its emperor and court lived in Kyoto, but the Kamakura shogunate, a warrior government based to the east, exercised much of the country’s practical political authority.
Kublai initially tried to bring Japan into relations with his court through diplomatic demands. In 1266, his government prepared letters for the Goryeo king and Japan’s ruler; the letter intended for Japan reached Dazaifu, in northern Kyushu, in 1268. It proposed official relations but also warned of consequences if Japan refused. The National Archives of Japan preserves and explains the surviving diplomatic record, including this communication carried through Korea. The invitation to official relations was accompanied by a warning.
Neither Kyoto’s imperial court nor the military government at Kamakura agreed to the relationship being demanded. Subsequent embassies also failed. Accepting the proposed exchange could suggest recognition of Kublai’s higher status within the diplomatic order he sought to extend beyond the continent. For Goryeo, the outcome had immediate costs. Its rulers were already constrained by Mongol authority and had to furnish ships, troops and maritime expertise for an assault on Japan. Korean involvement was shaped by imperial pressure as well as by the demands of warfare.
Kublai organised the first expedition in 1274 while the Yuan was still fighting to conquer southern China. The Southern Song dynasty survived until 1279, five years after the 1274 assault on Japan. Accounts that place the invasion after the completed conquest of China obscure that overlapping chronology. With the Southern Song defeated, Kublai could draw much more extensively upon southern Chinese personnel and ships for the second expedition in 1281.
The 1274 Invasion: Tsushima, Iki and Hakata Bay
The first fleet departed Korea in the autumn of 1274 with a mixed force of Mongol, Chinese and Korean troops, sailors and support workers. It comprised hundreds of ships and many thousands of people, though surviving totals vary. Modern summaries often quote 900 vessels or 40,000 people, but neither figure should be presented as an exact count of fighting soldiers. Some sources include crews and other personnel, while different narratives offer inconsistent estimates.
The ships made for Tsushima, between Korea and Japan, and then Iki, an island nearer the Kyushu mainland. Local Japanese garrisons fought but could not keep the invaders from advancing. Those assaults exposed the islands’ limited defenses, offered the attackers forward positions and alerted Kyushu to the threat. The fleet proceeded to Hakata Bay on the northwest coast of Kyushu, near Hakata, an important trading and administrative center.
At Hakata, warriors mobilised by the Kamakura shogunate met the invaders. The Yuan-led force employed bows, organised foot soldiers and cavalry, and explosive projectiles called tetsuhau in Japanese accounts. Popular comparisons between samurai seeking single combat and disciplined Mongol formations overstate the contrast. Japanese warriors could pursue personal recognition while still fighting alongside others and changing their tactics. The invading army was itself diverse, combining units and weapons associated with several military traditions.
A source close in time to the invasion is the Mōko Shūrai Ekotoba, a pictorial narrative ordered by the samurai Takezaki Suenaga. He took part in the fighting in 1274 and afterwards sought official credit for what he had done. The Museum of the Imperial Collections identifies the surviving handscrolls as a thirteenth-century national treasure. They preserve testimony close in time to the events, but they also reflect the version of his service that Suenaga wanted recognised. The existing sequence has been rearranged, and restorations added some figures later.
In the paintings Suenaga advances with a small group and is wounded; elsewhere foreign troops appear beside a rounded object often understood as an explosive projectile. The images are rare evidence of how combat was represented near the time, but their visual details cannot be treated as untouched photographs of the battlefield. Suenaga commissioned a record of his bravery and reward claim. Later repairs further complicate identification of what belongs to the original painting. Princeton’s annotated scroll project compares reconstructions and copies to explain these alterations.
Following the fighting around Hakata, the Yuan army pulled back. The reason remains contested. Historical narratives refer to worsening weather, changed winds and ships lost on the return journey. Bad weather could have compounded a decision to withdraw after costly resistance, without having forced the entire retreat by itself. There is weaker support for the specific story of a great nighttime typhoon destroying exactly a third of the fleet. In In Little Need of Divine Intervention, Thomas Conlan draws on Japanese records and the scrolls to restore the military role of the defenders. His argument does not rule out a storm; it challenges the claim that a storm alone defeated the invasion.
Seven Years of Preparation Changed the Next Campaign
The Kamakura government assumed Kublai would try again. It mobilised coastal guards and ordered stronger defenses along Hakata Bay. Beginning in 1276, landholders from different Kyushu provinces built sections of a stone-and-earth barrier known as the genkō bōrui or ishitsuiji. The defenses covered much of the shore most vulnerable to landing, although they were never an unbroken wall of identical construction along the whole bay.
That work changed what an invading army could do after reaching the beach. In 1274, attackers had been able to engage Japanese forces close to the shore. A raised barrier made it harder for mounted troops or infantry to move directly inland from landing craft and helped defenders assemble behind protected positions. Coastal guard obligations and locally mobilised warriors gave the works their military value. An archaeological and documentary report from Kyushu University examines the Hakozaki section and explains the construction duties imposed on landholders.
Kublai organised a much larger second expedition. After the Southern Song fell in 1279, Kublai could mobilise southern Chinese shipyards, supplies and troops as well as the forces available through Korea. The plan for 1281 divided the invasion into an eastern fleet from Goryeo and a southern fleet departing China. Commonly repeated totals are about 900 ships and 40,000 people for the eastern contingent and 3,500 vessels and 100,000 for the southern. These are transmitted orders and estimates, rather than verified inventories of every vessel and person that actually reached Japan.
The southern fleet carried fighting men and supplies for their continued maintenance. Accounts of tools, seed grain and provisions suggest preparations for a lengthy campaign and possibly the maintenance of troops on occupied land. Building and outfitting thousands of vessels demanded timber and labour, while success depended on the two fleets arriving in some degree of coordination. Their commanders faced an immense supply problem: transporting troops, animals, drinking water, food and equipment over open sea, then keeping them supplied during fighting along a defended coast.
The Second Invasion Began Before the Great Storm
The eastern fleet left Korea in 1281, struck Tsushima and Iki, and moved towards Hakata Bay. Unlike in 1274, the shore was fortified and guarded. Japanese warriors resisted from the defenses and launched attacks in small craft against Yuan vessels at anchor. The fighting ranged across the approaches to Kyushu and nearby islands, including Shiga Island, instead of taking the form of one continuous assault on a single beach.
The southern force made its voyage separately, and the two armies did not arrive together as intended. Delays and revised meeting points hindered coordination. A history of the campaign published by the Institute of Asian Underwater Archaeology follows the eastern and southern fleets through the fighting at Shiga Island and their movements into western waters. It draws on documentary accounts and evidence from Takashima, while recognising that exact dates differ depending on how medieval calendars and source traditions are interpreted.
By the campaign’s later stages, large numbers of Yuan vessels lay in the waters around northwestern Kyushu, notably Imari Bay and Takashima. Ships at anchor were exposed to Japanese raids, and their crews depended on safe approaches and secure supplies far from home. Crowded, shallow waters made manoeuvring heavy vessels difficult even in calm conditions. In a storm, ships driven into one another or ashore could suffer damage made worse by broken moorings and hulls. We cannot now locate every vessel at the moment the destructive winds arrived.
Japanese forces had already made the planned foothold on the mainland hard to obtain. The stone barrier did not enclose every possible landing area on Kyushu, and the Yuan did fight and land elsewhere. Along Hakata Bay, however, it blocked an otherwise accessible route onto the plain, while Japanese fighters harassed or boarded ships offshore. Japanese warriors had been fighting for weeks, supported in part by the fortifications built after 1274, before the storm struck.
What Happened to the Fleet in 1281?
During the summer of 1281 a violent storm struck the Yuan armada. Ships were wrecked, and troops lost their means of retreat. The event is usually placed in mid-August, but 15 August, often given as the exact date, should not be confused with an independently recorded modern-calendar observation. Japanese and Yuan writers used different traditions and calendars, and historians have converted the medieval dates in different ways.
Vessels broke up at anchor or on the coast. Soldiers drowned, while others were stranded on islands or ashore and remained vulnerable to Japanese attacks. Survivors were killed or taken prisoner. There is no way to reconstruct a precise total of deaths. Claims of 70,000 or 100,000 casualties, or exactly 4,000 ships destroyed, mix reported figures with later calculations. They convey how large the disaster was thought to be, not the results of an actual casualty census.
The ships were exposed to different levels of danger. Construction, hull form, loads, anchorage, water depth and commanders’ ability to order an organised departure could all affect whether a vessel survived. Wrecks linked with ships from southern China have been excavated, but that does not justify describing the whole Yuan armada as a fleet of unsuitable riverboats. It contained more than one kind of vessel, and a few preserved hulls cannot represent thousands of ships. Randall Sasaki’s 2009 shipbuilding study uses documents and archaeological finds to distinguish Chinese cargo vessels, other Chinese ship types and Korean landing craft.
By then the invasion was already struggling to achieve a secure landing. The storm destroyed ships that made it possible to move men and supplies, retreat or maintain command between separated forces. Stranded soldiers, however numerous, could not simply advance as a coordinated army towards Japan’s capital without transport, provisions and reliable leadership. It remains impossible to know whether the Yuan would have prevailed in calmer weather. The sources do show why losing their maritime support turned an exceptionally difficult campaign into a disaster from which it could not recover.
The Shipwrecks at Takashima
For centuries, accounts of the fleet’s destruction rested largely on chronicles and pictures. Fishermen near Takashima brought up pottery, weapons, anchors and other material from the seabed. Underwater investigations beginning in 1980 eventually established the remains of the 1281 invasion in those waters. The Takashima-Kōzaki area became Japan’s first nationally designated underwater historic site in 2012. Matsuura City’s archaeological summary describes that designation, the finds and a ship’s keel discovered in 2011.
The first recognisable shipwreck lay in water about 23 to 25 metres deep. A preserved section of keel measuring roughly 12 metres suggests the vessel was considerably longer, though no complete design can be recovered from one timber. Archaeologists have also found hull pieces, ceramics, lacquerware, arrows, iron helmets and other arms and equipment. These finds show what the fleet carried and provide physical evidence for maritime losses that medieval accounts generally express in enormous numbers.
Excavations have identified further vessels. After a second hull was found, investigators confirmed a third in October 2024. Matsuura City’s April 2026 report on Takashima Shipwreck No. 3 places it roughly 150 metres from shore in about 18 metres of water and about 50 metres south of the second wreck. Work in 2025 exposed keel timbers and their joints, together with ceramic and metal finds. The city cautions that the vessel is still incompletely understood. Identifying another wreck adds to the archaeological evidence without settling how many ships or men the invasion lost.
The Takashima wrecks have also been studied alongside evidence from other Yuan naval campaigns. Jun Kimura and colleagues discuss them in the peer-reviewed 2014 paper “Naval Battlefield Archaeology of the Lost Kublai Khan Fleets”, comparing the Japanese site with Vietnam’s Bạch Đằng River, where Yuan naval campaigns also left recoverable material traces.
The excavated remains show that ships associated with the Yuan invasion were lost near Takashima. They cannot yet separate losses caused by enemy action from those caused by wind, or identify the exact storm that damaged each hull. Condition of the timbers, sediment, weapons and datable objects can help narrow the possibilities, but archaeology cannot turn a medieval wreck into a precise meteorological observation by itself.
What Sediment Cores Say About the Divine Winds
The sediments around Kyushu provide a different kind of evidence. Jonathan Woodruff, Kinuyo Kanamaru, Shohini Kundu and Timothy Cook studied sediment at Lake Daija on Amakusa Island, south of the main invasion battles. Their work appeared online in December 2014 and in Geology in 2015, and the full research paper is available through Woodruff’s university site. The aim was to see whether the lake recorded exceptionally strong coastal storms around the invasion years.
Large waves can wash marine sand and other material across a barrier beach into a lagoon, leaving a trace in layers of lake sediment. The researchers compared deposits from Lake Daija with beach and freshwater catchment material, then applied X-ray imaging, strontium and inorganic-content analysis, radiocarbon dating and a statistical model for the age of each layer. Two unusually strong and closely spaced deposits date broadly to the late thirteenth century, close to the decades when the invasions took place.
The dating cannot securely match those two deposits to the individual years 1274 and 1281. Radiocarbon measurements have margins of uncertainty, and coastal overwash can arise from tsunamis as well as typhoons, although the historical and local setting makes severe storms the likelier explanation. Nor do the deposits give actual wind speeds at Hakata Bay or Takashima. By comparison with modern events, the authors suggest that the waves at Lake Daija might have required exceptionally intense storms, perhaps equivalent locally to category-three or stronger typhoons. Their study does not establish the tracks or peak intensity of particular medieval storms.
The sediment record also suggests more frequent extreme overwash at certain times in the surrounding centuries than during the recent period of instrumental observation. The team considered whether El Niño patterns, which can influence western Pacific cyclone tracks, might have contributed to that broader variation. This is a hypothesis about regional climate conditions over time. It does not demonstrate that an El Niño event drove a particular storm onto the Yuan fleet on a specified date.
The lake deposits give independent support for destructive storms in the late thirteenth century. They cannot determine why the first expedition withdrew in 1274 or whether either attack would have succeeded without bad weather. Military records describe decisions and fighting; sediments record physical processes near a coastline. The strongest account of the invasions draws upon both, while recognising that they address different parts of the history.
The Scrolls, Samurai Rewards and the Cost of Defense
Takezaki Suenaga’s scrolls also show how the invasions affected relations between warriors and the Kamakura government. The Kamakura government recognised and rewarded service by its warriors, often through land and privileges. Defending home territory did not, however, generate newly conquered estates to divide among victors. Suenaga’s illustrated account devotes considerable space to his efforts to have his fighting acknowledged, because securing that recognition was not automatic.
Coastal defense imposed continuing obligations on the shogunate’s landholding military supporters. It required men, horses, ships, arms, labour on the stone barrier and long periods of watch duty. Even after the Yuan disaster of 1281, the Kamakura authorities feared another invasion and maintained preparations. For warriors living in Kyushu, the cost of service continued after the enemy fleet had gone.
These campaigns added to the pressures facing the Kamakura regime, which fell in 1333, but its collapse cannot be explained solely by the invasions. Land disputes, political rivalries, changing relations with the Hōjō regents and wider discontent among warriors also contributed. A related Spoken Past history of the shogunate and imperial authority sets out the system within which the defense was organised. The invasions exposed a weakness in that system: prolonged demands for service were harder to meet when the government had few new lands with which to reward those who answered them.
How the Storms Became Divine Winds
Japanese shrines and Buddhist temples organised prayers and rituals while the invasions threatened the country. In later religious accounts, the fleets’ destruction became evidence that protecting deities had intervened. Hachiman, a deity strongly linked with warriors, was among the sacred powers associated with Japan’s defense. Such explanations tell us how religious communities interpreted the outcome. They should be read as historical expressions of belief, rather than independent reports of how the weather formed.
The word kamikaze joins kami, a deity or sacred presence, with kaze, meaning wind, and is commonly translated as “divine wind.” The expression and its associations were not created from nothing in 1274; wind, gods and sacred places were linked in earlier Japanese poetry and language. What the Mongol invasions added was a particularly influential interpretation of storms as divine protection against a foreign army.
Later generations continued to reshape the interpretation of the storms. Chronicles, paintings and monuments made the storms an important part of the memory of resistance to overseas conquest. Judith Vitale’s 2024 study of the historical writing of the invasions examines the ways early modern Japanese writers remembered the places, ancestors and events associated with the attacks. The familiar divine-wind narrative was reshaped over centuries before becoming part of the language of twentieth-century nationalism.
During the twentieth century, Japan’s military used the name kamikaze for suicide attacks against Allied ships, drawing on older ideas of sacrifice and national rescue. The label does not mean that the samurai of 1274 and 1281 fought in the same manner or shared the political beliefs of the pilots who followed them centuries later. In the 1920s or 1930s Uenaka Chokusai painted Divine Wind, showing Mongol ships caught in violent seas as Japanese fighters watch. The Honolulu Museum of Art’s object commentary explains the painting’s period and its relationship to the later nationalist importance of the medieval story.
The evidence shows why storms alone cannot explain both invasions. Written testimony and shipwrecks demonstrate the catastrophe of 1281, while geological work points to powerful storms near Japan in the late thirteenth century. The explanation for the 1274 withdrawal remains less certain, and both invasions were shaped by determined Japanese resistance and difficult Yuan logistics. Kublai continued considering a further attack after 1281, and preparations for one did not immediately cease, although no third invasion took place. The storms became the best-remembered part of the story because the invasions failed. That outcome also depended on the decisions of the commanders, the people mobilised to fight and build defenses, and the vulnerable fleets on which the Yuan army relied.
Discussion
Comments