THE AKKADIAN empire flourished on the banks of Euphrates in Iraq, from 2300 BC to 2200 BC. Though the reason for its sudden collapse has long puzzled archaeologists, Harvey Weiss and his colleagues at Yale University in USA now say it literally dried up and withered away.

The archaeologists base their claim on excavations in Akkadian sites in northern Mesopotamia, historical records of population movements preserved in clay tablets, and an analysis of soil moisture dating back to that era.

Though earlier explanations had put the blame on invading tribes or infighting among governors, the researchers say the empire's fall was triggered by a sudden shift in the climate. Unlike the southern Mesopotamians who had irrigation systems based on water from the Euphrates, the northerners depended on rainfed agriculture. A sudden severe drought robbed them of their lifeline -- wheat, barley and sheep (Science, Vol 261, No 5124).

The researchers argue food and water supplies in the south were strained to the point of collapse as the drought caused the northerners to migrate en masse to the south. The first signs of a drought -- fewer earthworm holes and wind-blown pellets and dust -- were dug up in Tell Leilen, a northern Akkadian site by Weiss' colleague, Marie-Agnes Courty, of the Centre National de Recherche Scientifique in Paris.

Data collected by the researchers suggest the region had become arid in 2200 BC and stayed dry for about 300 years. This is in tune with findings that suggest these sites were not occupied between 2200 BC and 1900 BC.

But what caused the dry spell is still not known. Courty's soil analysis suggests a volcanic eruption, though other scientists doubt if this could disturb the climate over such a large area for 300 years. Courty and Weiss also suspect massive warming of ocean currents could have disrupted local weather patterns.

Some archaeologists are, however, not convinced of the link between drought and decline of the empire. Robert Adams of the Washington-based Smithsonian Institution feels it is premature to extrapolate from soil moisture data drawn from one site. "The real problem," he explains, "is you can have tremendous local variability" in climate. Soil data from sites further north are needed to show whether they, too, suddenly dried up, he says.