No one in Japan felt the ground move.

That was what made it impossible. In the early hours of January 27, 1700, water rose on Japan's Pacific coast in ways that every coastal villager and every shogunate scribe recognized as the behavior of a 津波 (tsunami). At Kuwagasaki in present-day Miyako, waves crested at roughly four meters and tore apart thirteen houses; fires broke out and consumed twenty-one more. At Tsugaruishi, the water surged a full kilometer up the river from the coast, reaching the inland crossing at Kubota — farther than anyone had seen water travel inland without a parent earthquake. At Tanabe, wave heights reached between three and five meters at different points along the shore. At Miho, the village headman watched the water arrive with the characteristic behavior of tsunami flooding and immediately ordered his villagers to evacuate to high ground.

All of this was exactly what a tsunami looked like. But the ground had not shaken. Anywhere.

Edo-period Japan had a mature, sophisticated understanding of tsunamis. The word 津波 (tsunami), literally "harbor wave," described a phenomenon the coastal population knew well — sudden withdrawal of water, then catastrophic return, linked always to felt seismic shaking. The causation was understood: earthquakes under the sea moved water. The bureaucratic and scientific vocabulary of the Tokugawa period built this causation into the definition. A tsunami without an earthquake was not merely unusual. It was, in the language available to those scribes, almost a contradiction in terms.

The shogunate's administrative apparatus, one of the most thorough bureaucratic systems in pre-modern history, required the documentation of unusual events. Local officials filed reports. Regional administrators compiled them. The 幕府 (bakufu, shogunal government) maintained archives. And so the scribes did what they were trained to do: they wrote down what happened, and they wrote down that they did not understand it.

Some filed the event under "高潮" (takashio, high tide). Others used the phrase "高浪" (takanami, high waves) and noted, in language that has survived three centuries of archival storage, that no earthquake had preceded the event. The village headman of Miho wrote something particularly valuable: he described his own process of observation, recognition, and action. He saw the behavior. He made the call to evacuate. He recorded that he could not explain the absence of shaking. His document is a small masterpiece of empirical puzzlement.

The reports were filed. Damage was assessed, compensation was arranged, the rebuilding of Kuwagasaki proceeded according to the usual Edo-period rhythms of disaster and recovery. The mystery of the parentless wave was noted as anomalous and, in the absence of any explanation, set aside. The archives absorbed the documents and held them.

On the other side of the Pacific, no one had written anything down at all.

On the evening of January 26, 1700 — approximately 9:00 PM Pacific Time — the Cascadia subduction zone ruptured along a fault line stretching roughly 1,000 kilometers from what is now northern California to the middle of British Columbia. The rupture was a megathrust earthquake, magnitude estimated between 8.7 and 9.2 on the modern scale. The coastal land along the Pacific Northwest dropped one to two meters in seconds, instantly drowning vast forests of cedar and spruce in saltwater. A massive tsunami radiated outward in all directions.

The Indigenous peoples of the Pacific Northwest coast — the Huu-ay-aht, the Makah, the Quileute, and others — kept oral histories of a great catastrophic wave that had destroyed villages, driven survivors to high ground, and reshaped the shoreline overnight. These stories existed. They described exactly what had happened. But they were not written down in a form that 20th-century seismologists initially knew how to use.

In the 1980s and early 1990s, American geologist Brian Atwater began finding the physical evidence: "ghost forests" — dead cedar and spruce trees still standing in Pacific Northwest tidal flats, killed when the land dropped two meters in 1700 and saltwater drowned their roots. Their last growth rings dated to the fall of 1699. Atwater found buried sand sheets, the signature of tsunami inundation, layered above the drowned soil. The evidence said a massive earthquake had happened, probably sometime between late 1699 and mid-1700. But the date was not known.

Japanese seismologist 佐竹健治 (Satake Kenji) had a different kind of evidence: the shogunate archives.

Satake and his collaborators recognized that if the Cascadia earthquake had generated a tsunami large enough to damage Japan, the damage would have been recorded — and it was. Working through the regional reports, they found the accounts of Kuwagasaki, Tsugaruishi, Tanabe, Miho. They found the headman's puzzled document. They found the careful scribes who had written "high waves, cause unknown."

The physics of tsunami propagation across the Pacific is predictable. Given the wave heights recorded in Japan and the distance from the Cascadia zone, the travel time could be calculated. Working backward from the Japanese arrival times — recorded in shogunate documents using the traditional Japanese timekeeping of the period — Satake's team was able to pinpoint the earthquake to approximately 9:00 PM on January 26, 1700.

The result was published in the journal Nature in 1996. The shogunate scribes who had written "cause unknown" had, without knowing it, preserved a clock. Their frustration — their documented confusion, their careful notation of the inexplicable — was exactly what made the records scientifically useful. A scribe who had simply filed the event as unremarkable would have recorded less. A scribe who had invented a convenient earthquake would have falsified the record. The ones who wrote "we do not know" told the truth, and the truth survived.

The consequences of the 1996 paper were immediate and significant. Before Satake's work, many North American seismologists had doubted that the Cascadia subduction zone could produce a full-margin megaquake — the kind that kills tens of thousands and generates Pacific-wide tsunamis. The zone had been quiet in the historical record, and some researchers argued it might be "aseismic," slowly slipping without catastrophic rupture. The Japanese records ended that argument. The Cascadia zone could and had produced a magnitude-9 earthquake. It would do so again.

The Cascadia subduction zone ruptures on average roughly every 243 years, based on geological evidence counting past events over ten thousand years. It has now been 326 years since January 26, 1700.

The ghost forests still stand in the Pacific Northwest tidal flats. The dead trees killed when the coast dropped in 1700 are still there, slowly weathering, their last rings from the autumn before the wave. Across the Pacific, in archives that the Tokugawa bureaucracy built and maintained with the thoroughness that was both its genius and its rigidity, the documents that prove when those trees died are still there too — the scribes' puzzlement intact, still useful, still a clock.