About 34 million years ago, the boundary between the Eocene and Oligocene epochs, Earth's climate shifted dramatically from toasty temperatures in the Arctic Ocean and no ice at the poles to a rapid build-up of ice sheets in Antarctica. But little evidence exists for whether ice built up in the Northern Hemisphere during that same time.

Part of the problem is that the sediment cores obtained from deep-sea locations in the Arctic have proved difficult to date accurately, says palaeoceanographer Ian Harding of the National Oceanography Centre in Southampton, UK. So Harding's graduate student James Eldrett suggested cross-referencing microfossil data with the times of Earth's magnetic polarity reversals to date several cores collected from the Norwegian-Greenland Sea.

While perusing the original notes for one core, they noticed that a pebble found in Eocene-aged sediment had been dismissed as contamination from the drilling. When they re-examined the core, which had been in storage since 1993, they found numerous other 'dropstones' — fragments of rock carried out to the sea by ice — at different depths. The group had uncovered the first physical evidence that ice had been present in the Northern Hemisphere during the Eocene–Oligocene boundary, some 20 million years earlier than previously documented (see page 176).

The excitement about a few pebbles belied their potential importance in an ongoing controversy. In 2005, some of Harding's colleagues published a study (H. K. Coxall et al. Nature 433, 53–57; 2005) based on oxygen isotope calculations from Pacific Ocean sites — concluding that Arctic ice formation at the transition between the Eocene and Oligocene epochs was a real possibility. But without physical evidence, such studies were open to interpretation.

In the face of such scepticism, Eldrett and Harding decided to do a multidisciplinary study to develop a more reinforcing story. Their dropstone observations supported the presence of ice, but Harding wanted to know whether the ice came from glaciers or from sea ice. And, if glacial, from which land mass did the ice originate?

The characteristics of the dropstones and of the sediment grains in the ice provided clues to their origins. “A fairly rounded pebble has a huge gouge with parallel striations,” says Harding. “It is a fantastic example of glacial erosion and it alerted us that we needed to investigate further.” And by comparing grain-size fractions of the sediment, the team determined that the profiles matched grain sizes typically seen in icebergs, not sea ice. Also, scanning electron microscopy of tiny quartz grains from the core revealed surface textures that were “indicative of glacial processes, of being ground up in ice”, says Harding.

Add to that the types of minerals and species of dinoflagellate fossils that the researchers found in the core, and the team could make a robust case that the ancient glacier had grated its way across the eastern coast of Greenland, before breaking up and melting over the drilling site.

Harding says that their work seems to indicate the existence of at least some polar ice in a world with much higher temperatures and carbon dioxide levels than the preindustrial era. But he quickly cautions the data are “far too fragmentary” to reach a definitive conclusion. “We cannot know from this one core how much ice there was or how much it fluctuated,” he says. Now, says Harding, the challenge is to determine how much ice was present and how stable it was.