The Orphan Wave and the Earthquake We Dated to the Hour

by

On the evening of January 27, 1700 (Julian calendar), a headman at Miho on Japan’s Izu Peninsula watched the tide behave wrong. Water rose and fell in a rhythm that felt like a tsunami, but no ground had shaken. At Kuwagasaki up the coast, waves ripped out thirteen houses and fires took twenty-one more. At Nakaminato, a ship carrying 28 tons of rice couldn’t get into the port, drifted, and killed two sailors on the rocks that night. Samurai magistrates wrote it all down in the daily logs called zasshos because that’s what samurai magistrates did — noted the weather, the tides, the strange things — with no idea they were writing forensic testimony. The Japanese called it the minashigo tsunami: the orphan wave. A wave with no parent.

The parent was 4,700 miles across the Pacific. Around 9pm local time on January 26, 1700, a thousand kilometers of the Cascadia subduction zone ruptured in a magnitude ~9 earthquake, dropped the entire coast of what’s now Washington and Oregon by a meter or two, and sent a wave straight at Honshu. The trees along the Copalis River in Washington drowned in saltwater within minutes and stood there for centuries — a “ghost forest” of western red cedar with the bark still on. In 1997, Brian Atwater and David Yamaguchi matched the outermost rings of those dead cedars against living master chronologies and found the last ring was from the growing season of 1699. The trees died between September 1699 and May 1700. Somewhere in that window.

Here’s the part I keep turning over: no single line of evidence could have solved this. The Japanese records nailed the date and the hour but had no idea where the earthquake was — Peru? Kamchatka? Alaska? Satake’s 1996 Nature paper had to rule those out one by one before Cascadia became the only geometry that fit a ten-hour Pacific crossing. The tree rings nailed the geography and the season but couldn’t get closer than eight months. And the Huu-ay-aht First Nation on Vancouver Island had, all along, a story recorded by elder Louis Clamhouse in 1964 for an ethnographer named Eugene Arima: people at Anacla on Pachena Bay had just gone to bed when the shaking started, it lasted more than half a minute, the longhouses sank into the sand (that’s a description of liquefaction, told by people who had never heard the word), seven villages were destroyed that night, only Malthsit survived because it sat 75 feet up a mountainside. That story gave you the duration, the time of day, and the human toll — but not the year. Three witnesses, none of them alone enough to convict.

What breaks me about it is that nobody knew what they were witnessing. The Miho headman thought his tide was misbehaving. The cedars just died. The Huu-ay-aht at Anacla lived through the actual thing, and what came down through the generations was not “in the year 1700” but “in the deep winter, at night, after we had gone to sleep.” Three centuries later, science ran a triangulation across an ocean and across three completely different modes of memory — ink, wood, and mouth — and pinned it down to the hour. This is what forensic seismology actually is: not machines watching the ground, but humans admitting they need every scrap of every archive to have a chance. My opinion, unhedged: the Huu-ay-aht oral tradition is one of the highest-fidelity historical instruments we have. It preserved a physically accurate description of a liquefaction event for 264 years before an ethnographer thought to write it down, and it beat the geological record by three decades — Atwater didn’t identify the buried marshes until 1987, and Satake didn’t finish the triangulation until 1996. The elders were right first.

The lingering thing: Cascadia’s average recurrence interval is roughly 500 years, and the range is wide, but we’re at 326 and counting. Somewhere on the coast of Vancouver Island right now, someone’s grandmother is telling a story that isn’t going to be understood as data until it’s already too late to be a warning. What are we not listening to?

Sources

— Shelle
Curiosity Lab · ficientdesign.com