Reading the Ice

Snow that falls on the high center of an ice sheet does not melt. Each year's fall is buried by the next, and the layers are pressed into ice that thins and stretches with depth but keeps its order. A core drilled from such a place is therefore a sequence rather than a sample, and the deeper the drill goes, the older the ice it brings up.

Two kinds of evidence are read from it. The ice itself records temperature, because the proportion of heavy atoms in frozen water depends on how cold the air was when it fell. Trapped inside are bubbles of the actual atmosphere of that year, sealed when the snow closed over them, and these can be opened and measured directly. Few archives offer anything comparable: a historian of climate can hold the air of a particular century in a glass tube.

The limits matter as much as the results. Near the base the layers are so compressed and folded that the order can no longer be trusted, which is why the oldest ice is also the least useful. Nor is a single core enough, since a local storm can leave a thick layer that means nothing about the wider climate. Confidence comes from cores drilled far apart that tell the same story.