Before we walked the earth, our planet was subject to abrupt short-lived climate swings. Richard Alley believes that this pattern could begin again.

Climate changes, and humans adapt. Through the millennia of human history, we have matched our rhythms to those of the Earth. But recent evidence from ice cores and elsewhere has shown that we have developed agriculture and industry during an unusual period of stable climate. Growing scientific knowledge suggests that human actions may affect whether or not the climate returns to its "normal" behaviour of wild fluctuation.

Climate history is written in sediments and other detritus of the past. The first European scientists to study the history of twigs, seeds and pollen in lake-bed mud learned that the end of the most recent Ice Age, about 12,000 years ago, was not smooth. Instead, swings between colder and warmer conditions forced tundra plants to migrate rapidly across the continent.

Those early studies did not reveal just how rapidly the climate changed, however, and it remained for ice-coring projects of the mid-1990s to show that changes of 8C or more have occurred repeatedly in about a decade. More recent studies have shown some of the reasons why those climate changes occurred and suggest that future changes depend in part on us.

³ÉÈËVRÊÓƵ

Central to these studies are the ice cores from Greenland, Antarctica and smaller mountain glaciers and ice caps scattered across the globe, which provide some of the finest histories of climate change. These ice cores can be dated accurately and provide information on changes in temperature and snowfall on the ice sheet, changes in wind-blown materials from well beyond the ice and changes in the gaseous composition of the atmosphere.

The great ice sheets of Greenland and Antarctica are about 3km thick in their centres and have been accumulating snow for hundreds of thousands of years. Gravity causes fallen snow to be squeezed to ice and to flow from central regions of an ice sheet to the coast, where the ice melts or breaks off to form icebergs. This flow of ice away from the centre stretches and thins layers, in the same way that a blob of batter spreads and thins to form a pancake. Deeper, older layers of an ice sheet have spread and thinned more than younger, shallower layers, so an ice core provides both a brilliantly clear view of recent events and a fuzzier view of the more distant past.

³ÉÈËVRÊÓƵ

Two of the most successful ice-coring projects were the coordinated efforts that collected parallel cores through the centre of the Greenland ice sheet between 1989 and 1993: the primarily European Greenland Ice Core Project (Grip) and the primarily US Greenland Ice Sheet Project 2 (Gisp2). These teams were transported by ski-equipped aircraft and supported mainly by European and US government funding agencies.

Dozens of scientists and drillers spent months each summer in the thin, subfreezing air of the Greenland Summit, collecting and analysing kilometres of core. Science labs carved into the snow on top of the ice sheet allowed continuous, rapid analysis of the cores, which were subsequently shipped home, still frozen, for ongoing study.

These cores - cylinders of ice about 10cm across in 1m long sections - were analysed electrically, chemically, isotopically and physically. Differences between summer and winter snow allow us to count annual layers like tree rings and checks against the ash from historically dated volcanoes show that such counting is fairly accurate. The varying thicknesses of the annual layers show the history of snowfall, while the dust, sea-salt, forest-fire smoke and other materials in the ice reveal the history of wind-blown materials. Bubbles trapped in snow and turned to ice contain samples of old air, which are preserved essentially unchanged until broken and analysed in laboratories.

This evidence from the ice cores shows that climate changes have been huge. The coldest part of the Ice Age on the surface of Greenland was more than 20C colder than recent temperatures, with only about 20 per cent of modern snowfall (it snows less in colder conditions) and between ten and 100 times as much windblown dust and sea salt.

Greenhouse gases, including carbon dioxide and nitrous oxide, were reduced in the atmosphere during cold times. Swamp-gas methane, another greenhouse gas, was also much lower then, showing that the ice-age world had reduced wetlands.

³ÉÈËVRÊÓƵ

More remarkably, in the 110,000-plus years of reliable ice-core record from Greenland, the few thousand years of human agriculture and industry are climatically the most settled years. The rest of the record is dominated by large, abrupt and widespread climate jumps, often about one-third to one-half of the entire difference between ice-age and recent conditions. They affected much of the Earth and lasted ten years or less. Some happened in a single year.

Climate records from numerous regions confirm this picture: ocean-core histories of windiness off the coast of Venezuela; of ocean oxygenation and temperature off the coast of California; and of windblown dust off Arabia all show the same patterns observed in Greenland ice. Much of the world seems to have jumped with Greenland between cold, dry, windy conditions and warm, wet, quiet times.

Ocean-sediment records, physical studies and model simulations all link the climate jumps to switches in the large-scale circulation of the ocean. Today, warm, salty waters flow far to the north along the surface of the Atlantic, giving up their heat to warm northern Europe before sinking to the deep ocean, flowing south, and rising in other parts of the world ocean before returning. Were the north Atlantic waters to become fresher for some reason, the water might freeze on the surface before becoming dense enough to sink, chilling European regions, weakening the monsoon rains of Africa and causing other changes around the globe. While some of the large coolings of the past have been linked to huge floods from ice-age, ice-dammed lakes that freshened the Atlantic, other coolings have not yet been explained. But, while other regions were probably involved, the evidence suggests that the north Atlantic played a key role.

³ÉÈËVRÊÓƵ

Our species is highly likely to continue increasing the greenhouse-gas concentration of the atmosphere, warming the world. Many models project that this warming will freshen the north Atlantic by melting some - or all - of the Greenland ice sheet and by increasing precipitation at high latitudes. This in turn could slow the ocean currents that warm northern Europe. Were this to happen, global warming could, ironically, end up freezing some regions, followed perhaps by an abrupt warming should the ocean circulation re-establish itself.

At least one model suggests that slowing down the rate at which we burn fossil fuels could allow the modern ocean currents to continue but that rapid burning of our fossil fuels could cause major changes to the oceans and atmosphere.

Concern about the prospects of rapid climate change has led to new coring projects in Greenland and on high-altitude tropical glaciers, in oceans and lakes, trees and cave formations, in a search for better histories. Many of us in the ice-core field have turned our attention southward, in search of Antarctic records.

Yet it is still too soon to make predictions. Future climate changes could be smooth and slow. But the ice cores and other records show clearly that climate "switches" can happen and are capable of throwing the earth's system of atmosphere, ocean, ice and living things into a new mode of operation in a decade or less. It is essential that we learn where these switches are before we accidentally blunder into one.

³ÉÈËVRÊÓƵ

Richard Alley is Evan Pugh professor at the Environment Institute and department of geosciences, Pennsylvania State University, United States. His book The Two-Mile Time Machine is published by Princeton University Press, Pounds 15.95.