The pebble's story

Seven years in the ice

29 October 2026 · Griff

I told Beat the ice sheets had been running seven years off, and the sun fixed it. She asked who counts ice sheets. Her dog gave the phone a look. Fair question. Fair look.

Start with trees, because the ice needed them.

In 1901 an astronomer at the Lowell Observatory in Arizona, A. E. Douglass, wanted to know if sunspots change the weather. He reasoned that trees at the dry edge of a forest might write each year's rain into the width of a ring. He started measuring rings in 1904 and found that trees across a whole region carried the same pattern of wide and thin years. Match the pattern from a living tree to an old beam to an older one, and you can count back year by year. That's crossdating (Laboratory of Tree-Ring Research).

He never really nailed the sunspots. He got a whole science instead. Not a bad consolation prize.

Ice works on the same idea, only colder. Snow falls on Greenland and Antarctica and never melts. Each year gets squashed under the next. Summer snow and winter snow differ in their chemistry: dust peaks in spring, sea salt in winter, the oxygen isotopes swing with temperature. So every year leaves a layer. People drill cores up to about three kilometres down and count those layers, metre after metre, under lights, in freezers.

A lot of them are Danes. Copenhagen ice work goes back to Willi Dansgaard in the 1950s, and today it lives at the Niels Bohr Institute, which built the Greenland Ice Core Chronology from counted layers. The Swiss came at it from Bern. Patient people. Cold hands.

The weak spot is plain. If you miss a thin layer, or count a messy one twice, every date below it slides. And nothing outside the ice could tell you.

Then, in 2012, Fusa Miyake and colleagues measured carbon-14 in the yearly rings of two Japanese cedars and found a sharp jump between AD 774 and 775. About twenty times a normal solar wobble (Nature, 2012). The best explanation now is an extreme solar particle storm. It shows up in every tree that was alive that year, wherever it grew. Tree rings are counted exactly, so that spike is a fixed date. A timestamp from the sun.

The same storm made extra beryllium-10, which falls out of the air onto the ice within a year or two. In 2015 Michael Sigl and colleagues, including Danes from the Niels Bohr Institute and Swiss from Bern, measured it year by year in three Greenland cores and one from West Antarctica. The peak was there in both ice sheets. The old ice dates put it at AD 768 (Nature, 2015).

Seven years. A second spike, the 993 to 994 event, came out about seven years off as well. The ice had lost a few years somewhere in the counting, and nobody could have known.

So they pinned the ice to the trees and rebuilt the timescales. Once the two agreed, big volcanic eruptions in the ice landed just before the cooling in the tree rings instead of after it. Cause before effect, which is the order you want. A separate study using chlorine-36 as well pointed to the storm being solar, and bigger than anything satellites have seen (Mekhaldi et al., 2015).

That's the same shape I keep finding. A clay clock anchored by a flood. Norse woodwork in Newfoundland dated by another storm. Now the ice, counted by hand for decades, put right by one loud year.

The counting was good work. It just needed one day that shouted.