Nan Madol

A city you can date by its coral

Much of Nan Madol is built of dead coral. That is how we can date it. When a coral head dies, the uranium it took up from seawater begins to decay into thorium at a known rate, so a sample from the fill of a wall records, often to within a few decades, when that coral died before it was built into the islet. In 2024 a team including Chuan-Chou Shen, Felicia Beardsley and Mark McCoy dated 167 coral samples from 18 islets this way, and the result moved the beginning of the city back by roughly two and a half centuries.

The islets themselves offer no farmland and no fresh water of their own.

What is actually there

Off the south-east coast of the island of Pohnpei, in the Federated States of Micronesia, artificial islets sit on a shallow reef, laced together by tidal canals. The count is usually given as 92, but UNESCO's own text says more than a hundred, and the listed property covers about 76.7 hectares. It was the ceremonial and political centre of the Saudeleur dynasty, the line of rulers remembered in oral history as having unified Pohnpei, and it has been called the Venice of the Pacific for as long as outsiders have been describing it.

The islets are walled with columnar basalt: naturally hexagonal rock columns, formed when lava cooled, stacked in a technique that resembles log construction more than conventional masonry, with no mortar. The largest structure, Nandauwas, is the royal mortuary islet, with walls about 7.5 metres high enclosing a central tomb, and measuring roughly 80 by 60 metres. Geochemical sourcing with portable X-ray fluorescence matched the basalt to natural sources on the far side of the island, formed one to eight million years ago, so the stone did not come from next door.

How old, and how the answer moved

Oral history, counted back by generations, put the start of the Saudeleur dynasty at around 1160 CE, and earlier dating placed the founding of Nan Madol around 1300.

2016. Mark McCoy, Helen Alderson, Richard Hemi, Hai Cheng and R. Lawrence Edwards dated coral from the Nandauwas tomb in Quaternary Research. They narrowed the founding to a 20-year window, 1180 to 1200 CE: by 1180 stones were being brought for the tomb, and by 1200 the vault held its first burial, the island's chief. They called it the earliest direct evidence of monument building at Nan Madol.

2024. Chuan-Chou Shen, Felicia Beardsley and colleagues, with McCoy as the senior author, published a far larger dataset in PNAS Nexus: 167 coral samples from 18 islets plus four published dates, and 18 charcoal samples for radiocarbon. They found two building phases. The first ran from about 930 to 1130 CE. After a gap of about forty years the second ran from the late twelfth century to about 1411 to 1425. The authors say this is centuries earlier than previously believed. They read a cluster of coral dates around 1055 to 1075 as the mark of a new ruling class, and note that in the late phase about 37 to 38 per cent of dated corals come from seawall compounds, against 7 per cent in the early phase. They connect this to rising sea level from the island's subsidence and to swings in El Niño conditions, and suggest the site's history is one of repeated investment in protection, not early abandonment.

The caveats are in the paper. The authors say it is difficult to establish a direct statistical link between climate and construction, that the climate evidence is largely qualitative and comes from records outside the site, that some corals may be old rubble or later repairs and so may not date a single building episode, that charcoal of unknown plant species dates poorly in the Pacific, and that only 18 islets were sampled. In the late phase the El Niño record and the coral use diverge, so they do not claim El Niño as the main explanation. None of this needed a lost civilisation. It needed better instruments.

What is still open