The rumor that Koreans were poisoning wells and setting fires after Japan's 1923 Great Kanto earthquake was a false, baseless hoax that triggered the massacre of an estimated 6,000 people
Where the evidence lands: ContradictedThe claim, as it was spread in September 1923, held that Koreans in the Kanto region were exploiting the disaster: poisoning the wells that survivors depended on, setting the fires that were consuming the city, looting ruined homes, and organizing to rise up against Japanese rule. The rumor cast an entire minority as a fifth column that had to be hunted down for public safety.
Believed by: The rumor was believed by large numbers of ordinary Japanese in the immediate panic and was propagated by parts of the police, press, and government. Among historians today there is no dispute that the well-poisoning claim was false; the massacre it caused is established, while a fringe of nationalist denialism in Japan continues to minimize the toll or dispute that it happened at all.
The full story
What happened
At 11:58 a.m. on 1 September 1923, a magnitude-7.9 earthquake struck the Kanto plain, the region around Tokyo and Yokohama. The quake itself was deadly, but it was the fires that followed, feeding on densely packed wooden houses and the charcoal braziers lit for the midday meal, that made the disaster one of the worst in Japanese history. Wind-driven firestorms, including a firewhirl that killed tens of thousands of people sheltering at a single former army depot, pushed the death toll to roughly 105,000.
Into that vacuum of information and order came a rumor. Within hours, word spread through Yokohama and Kawasaki and then into Tokyo that Koreans, a colonized minority resented and feared after more than a decade of Japanese rule over Korea, were poisoning the wells, setting the fires, looting the ruins, and organizing an uprising. It was a false rumor from the start. But it arrived in a city with no reliable news, a terrified population, and a reservoir of colonial prejudice ready to receive it.
What that rumor set off was not a riot but a hunt. Over the following weeks, neighborhood vigilante bands, police, and imperial soldiers killed an estimated 6,000 people, the overwhelming majority of them Korean. This file is about that rumor and the atrocity it caused: the claim that Koreans were poisoning wells was baseless, and the killing that followed is documented fact.
The rumor was false, and known to be
There was no well-poisoning. The claim was never substantiated in a single instance, and its central piece of apparent evidence has a complete and mundane explanation. The earthquake and the fires shook debris, plaster, dust, and ash into the region's open wells; water that had been clear turned cloudy overnight. To people already told that Koreans were sabotaging the city, murky water looked like proof. It was nothing of the kind. It was what an earthquake does to a well.
Some officials knew this and said so. The police chief in Tsurumi publicly drank water from a well to demonstrate that it had not been poisoned, a direct attempt to kill the rumor with a visible test. It was not an isolated doubt: the authorities' own later reporting on the events conceded, in a confidential appendix, that the stories of Korean sabotage had been unfounded. The arson claim collapses just as fast. A huge quake striking a wooden city at cooking time, followed by firestorms, needs no arsonists to explain the flames.
The tragedy is that the correction never caught up with the panic, in large part because the state had helped spread the panic in the first place.
A police chief drank from a well in public to prove it was not poisoned. The rumor was already recognized as false while the killing it caused was still under way.
How a rumor became a massacre
On 2 September the government declared martial law. Instead of suppressing the rumor, parts of the police and the Home Ministry relayed it, circulating warnings that Koreans posed a threat. That is the hinge of the whole episode. When the story came stamped with official authority, ordinary people stopped hearing a wild tale and started hearing what the state apparently believed, and they acted on it.
Neighborhoods organized into jikeidan, self-defense committees numbering in the thousands across the region and frequently led by military reservists. They armed themselves with swords, bamboo spears, hooks, and iron bars, threw up checkpoints, and forced passers-by to pronounce Japanese words that Korean speakers were thought to struggle with. Those who failed the test were beaten and killed on the spot. The method caught others in its net: Chinese laborers, and Japanese from regions whose accents sounded wrong at a checkpoint.
The violence was not confined to civilians. Historians document that police organized and armed some of the vigilante bands and that imperial army units took part directly, in places turning machine guns on Koreans. Under the same cover, the state settled scores with the Japanese left: in the Amakasu Incidenton 16 September, military police murdered the anarchists Sakae Osugi and Ito Noe along with Osugi's young nephew. A false rumor about one minority had opened a season of killing that reached well beyond it.
The toll, and the long denial
Almost as soon as the killing stopped, the state began to shrink it. Contemporary official accounts acknowledged only a handful of Korean dead, and a later police estimate reached about 231. Independent investigation told a very different story. The Korean community's own count and the work of Japanese intellectuals, among them the Tokyo Imperial University professor Yoshino Sakuzo, who publicly condemned the authorities, documented thousands of victims. The figure historians most often cite is around 6,000, with some estimates higher still.
That gap between roughly 231 and roughly 6,000 is not a genuine historical uncertainty about whether the massacre happened. It is the measure of a state that declined to count its own victims. And the reluctance did not end in the 1920s. No Japanese government has ever conducted a full accounting of the killings or of officialdom's role in them.
The denial reaches the present. On 30 August 2023, days before the centenary, Japan's chief cabinet secretary said the government had found no recordsthat would let it confirm the killings took place, a formulation critics noted was at odds with the documentary record and even with the official's own earlier acknowledgment. A fringe of nationalist writing goes further, disputing the toll or the massacre itself.
The distance between the official figure and the historians' estimate is not doubt about the atrocity. It is the shape of a century of refusing to look at it.
Why it still matters
The Kanto massacre is a case study in how a false rumor kills. Every element that made it lethal is legible: a catastrophe that stripped away information and order, a minority already cast as suspect by years of colonial rule, an ambiguous physical detail that seemed to confirm the worst, and, decisively, a state that amplified the story instead of correcting it. None of the killers needed a genuine threat. They needed permission, and the rumor, dressed in official authority, supplied it.
Reporting it honestly means holding the frame steady. The claim that Koreans poisoned wells is not a contested historical question with two respectable sides; it is a debunked hoax, recognized as false while it was still being told. The people who died were not perpetrators of anything. They were the victims of the rumor, hunted at checkpoints over the way they spoke.
What keeps the story urgent is the unfinished reckoning. The facts are established, but the acknowledgment is not, and a settled atrocity that the state still will not fully name remains vulnerable to the same denial that let it happen. Grassroots memorials, survivor testimony, and scholars in Japan and Korea carry the memory that officialdom has not. Repeating the truth plainly, that the well-poisoning claim was a lie and that thousands were murdered because of it, is part of keeping it from being erased.
What's still unexplained
- The exact death toll has never been settled. Official figures put it in the low hundreds, while independent and scholarly estimates cluster around 6,000 and range higher, a gap that reflects the state's failure to count its victims rather than genuine uncertainty about whether the massacre occurred.
- The full extent of state responsibility is still not officially acknowledged. Police organized vigilante bands and army units took part directly, yet no Japanese government has conducted a full accounting, and the question of command responsibility remains formally unaddressed a century later.
- Denialism persists at the highest levels. As recently as 2023 the government said it could find no records confirming the killings, and a nationalist current in Japan continues to minimize or dispute the massacre, which keeps a settled historical atrocity politically contested.
- Memory is unevenly preserved. Grassroots memorials, survivor testimony collections, and Korean and Japanese scholars keep the record alive, but without state recognition the history depends on civil society, and each anniversary reopens the question of whether Japan will formally confront it.
Point by point
The claim: Koreans really were poisoning the wells, and the cloudy water proved it.
What the record shows: This is false, and the cause of the murky water is well understood. The earthquake and the firestorms shook loose debris, plaster, dust, and ash that fouled open wells across the region. Frightened survivors, primed by colonial prejudice, read contamination as poison. No case of well-poisoning was ever substantiated; officials who tested the claim, including a police chief in Tsurumi who drank from a well in public, found nothing, and internal police reporting later admitted in a confidential appendix that the rumors were unfounded.
The claim: Koreans were setting the fires that destroyed Tokyo and Yokohama.
What the record shows: Also false. The fires had an obvious and sufficient natural cause: a massive earthquake striking a city of densely packed wooden buildings just as midday cooking fires and braziers were lit, followed by wind-driven firestorms including a firewhirl that killed tens of thousands at a single site. Arson by Koreans was not the source of the conflagration; it was a rumor grafted onto a catastrophe that needed no human villain.
The claim: The rumor was a spontaneous, understandable reaction to real danger.
What the record shows: It was neither well-founded nor merely spontaneous. The claim tapped a decade of anti-Korean colonial anxiety after the 1910 annexation and the 1919 independence uprising, and it was actively spread through official channels: police and Home Ministry communications relayed warnings about Koreans, and some newspapers printed the rumor as fact. State amplification, not just street panic, turned a false story into a license to kill.
The claim: Only a small number of people died in the disorder.
What the record shows: This reflects the state's own minimizing rather than the record. Contemporary official figures acknowledged only a handful of Korean deaths, and a later police estimate reached about 231. Independent investigation, including work by the Korean community and Japanese scholars and intellectuals such as Yoshino Sakuzo, documented a far higher toll; the estimate most widely cited by historians is around 6,000, with some ranging higher.
The claim: This was mob violence that the authorities tried to stop.
What the record shows: The authorities were not simply bystanders. Vigilante jikeidan carried out much of the killing, but police organized and armed some of these bands, and imperial army units took part directly, in places using machine guns against Koreans. Historians note that soldiers actively participated and that martial-law forces were implicated, which is one reason the state has been reluctant to fully investigate its own role.
The claim: Only Koreans were killed, so this was narrowly about one rumor.
What the record shows: The victims were overwhelmingly Korean, but not only Korean. Chinese laborers were killed, Japanese from regions with unfamiliar accents were murdered when they failed shibboleth tests at checkpoints, and Japanese socialists, anarchists, and labor organizers were killed under the same cover, as in the Amakasu Incident. The false rumor opened a broader season of state and vigilante violence against anyone marked as an outsider or a threat.
The claim: Because it happened a century ago, the facts are hazy and disputed.
What the record shows: The core facts are not hazy. Survivor testimony, contemporary journalism, diplomatic records, police documents, and decades of scholarship in Japan and Korea establish that the well-poisoning claim was false and that a mass killing followed. What remains contested is the precise death toll and, above all, the degree of state responsibility, a gap kept open less by missing evidence than by an official reluctance to confront it that persists to the present.
Other readings
Angles that don't fit neatly into the claim or its rebuttal, laid out and weighed, not endorsed.
The murky-water explanation
The single detail that seemed to give the rumor substance, cloudy well water, has a plain and complete explanation. A magnitude-7.9 earthquake and the firestorms it set off shook debris, plaster, soil, dust, and ash into open wells across the region. Water that had been clear turned turbid overnight, and in a climate of fear that ordinary physical consequence of the disaster was misread as deliberate poisoning. Understanding this is not a defense of the rumor; it shows precisely how a false claim can feel evidenced. There was something real to point at, and prejudice supplied the sinister interpretation.
Why the denial persists
A century on, the live dispute is not about the facts of the killing but about whether the Japanese state will own them. Because police organized some vigilante groups and soldiers took part directly, full acknowledgment implicates the government itself, which helps explain a pattern of official minimizing that runs from the low contemporary body counts to the 2023 statement that no confirming records could be found. This angle does not reopen the question of whether the massacre happened, which is settled; it addresses why a settled atrocity remains politically unfinished business.
Timeline
- 1910Japan formally annexes Korea, beginning a colonial rule marked by land dispossession, forced labor migration, and the harsh suppression of the 1919 March First independence movement. By 1923 tens of thousands of Koreans live and work in the Japanese home islands, many as low-paid laborers, amid widespread official suspicion and popular prejudice.
- 1923-09-01At 11:58 a.m. a magnitude-7.9 earthquake devastates Tokyo, Yokohama, and the surrounding Kanto region. Firestorms fed by collapsed wooden houses and lunchtime cooking flames sweep the cities; the disaster ultimately kills around 105,000 people and leaves millions homeless.
- 1923-09-01Within hours, rumors begin circulating in the Yokohama and Kawasaki areas that Koreans are poisoning wells, committing arson, and planning riots. Well water had turned murky with debris, dust, and ash from the collapse; in the panic this ordinary consequence of the quake is reinterpreted as sabotage.
- 1923-09-02The government declares martial law over the stricken region. Rather than quashing the rumor, parts of the police and Home Ministry relay it: warnings that Koreans are a threat go out through official channels, lending the false story the authority of the state and helping organize the response around it.
- 1923-09Neighborhoods form jikeidan, self-defense vigilance committees, numbering in the thousands across the region and often led by military reservists or local notables. Armed with swords, bamboo spears, hooks, and iron bars, they set up checkpoints, demand that passers-by pronounce Japanese shibboleths, and kill those judged to be Korean.
- 1923-09Officials attempt in places to counter the rumor: the police chief in Tsurumi publicly drinks water from a well to demonstrate that it has not been poisoned. Such gestures do little against a panic the authorities themselves had amplified, and the killing continues for roughly three weeks.
- 1923-09-16In the Amakasu Incident, military police murder the prominent anarchists Sakae Osugi and Ito Noe and Osugi's young nephew, one strand of a wider wave in which Japanese socialists, communists, and labor organizers are killed under the cover of the disorder, showing the violence extended beyond the Korean target of the original rumor.
- 1923-10As order returns, the state moves to contain the story rather than confront it. Contemporary official tallies acknowledge only a handful of Korean deaths; a later police estimate puts the figure around 231, far below the roughly 6,000 that scholars and the Korean community would document.
- 2023-08-30On the eve of the massacre's centenary, Japan's chief cabinet secretary Hirokazu Matsuno says the government has found no records that allow it to state that the killings took place, drawing criticism that a century on the state is still declining to acknowledge an atrocity that is thoroughly documented.
Contradicted. The claim at the center of this file, that Korean residents were poisoning wells, setting fires, and arming for a rebellion in the hours after the 1 September 1923 Great Kanto earthquake, is false. It was a panic rumor with no factual basis, and it was recognized as baseless at the time: a police chief in Tsurumi publicly drank from a well to show the water was safe, and internal police reports later conceded in a confidential appendix that the stories were untrue. What the rumor produced was real and catastrophic. Over roughly three weeks, vigilante bands, police, and imperial soldiers hunted down and killed an estimated 6,000 people, overwhelmingly ethnic Koreans, along with some Chinese laborers, some Japanese mistaken for Koreans, and Japanese leftists killed under the same cover. This file rates the well-poisoning claim debunked and reports the killing as documented history. The one genuinely open matter is not whether the rumor was false, which is settled, but the exact death toll and the Japanese state's long refusal to fully investigate or acknowledge its role, a denialism that continues into official statements made in 2023.
Reviewed by The Conspiratory Editors · Last reviewed July 19, 2026 · How we rate
Sources
- 1.'It hurts my heart': Japan's Kanto massacre, 100 years on, France 24 / AFP (2023)
- 2.Japan Marks 100th Anniversary of the 1923 Kanto Massacre, The Diplomat (2023)
- 3.For Koreans in Japan, this little-known massacre still carries weight, The Christian Science Monitor (2023)
- 4.The Great Kanto Earthquake and the Massacre of Koreans: State Responsibility and Popular Responsibility, Harvard-Yenching Institute
- 5.The Massacre of Koreans in Yokohama in the Aftermath of the Great Kanto Earthquake of 1923, Kenji Hasegawa, Monumenta Nipponica (Sophia University) (2020)
- 6.The Kanto Massacre: A Story of Disinformation and Denialism, Unseen Japan (2023)
- 7.Xenophobia and misinformation: Why Japan's 1923 Great Kanto Earthquake led to a Korean massacre, Milwaukee Independent (2023)
- 8.Behind the Accounts of the Great Kanto Earthquake of 1923, Brown University Library (Center for Digital Scholarship)
- 9.Kanto Massacre, Wikipedia
Help us investigate
This is a living case file. If you spot an error or know evidence we missed, tell us, and weigh in on where you land.
Where do you land?
Cast your read on this one.
Comments
Add your take. Comments are read and approved by a human before they appear, so keep it on topic and civil. Please do not accuse named, living people of crimes.