Climatologists puzzling over what Earth might be like after global warming have long had a problem. Computer models have not been able to simulate the conditions that deep-ocean sediments suggest prevailed during similar ancient warm periods on Earth. Sediments seem to record a greenhouse climate with cool tropical oceans; models imply that the ocean surfaces were rather warmer.

Now researchers have re-examined the sediments and found that, for once, the data, not the models, were at fault. The resolution of this paradox implies that global warming will make the tropics hotter.

The tropics are home to many developing countries - the very places particularly vulnerable to climate change, lacking the resources to adapt. Here, ocean temperatures determine monsoons and wind speeds. Temperature differences between the tropical oceans and the high latitudes are also central to our understanding of the climate system.

Past ocean temperatures are usually determined by analysing the shells of foraminifera, tiny organisms ubiquitous in the sea. Just as the broken watch on a murder victim's wrist in an Agatha Christie novel preserves the time of the crime for posterity, foraminifera preserve temperature. Like Hercule Poirot, scientists using the shells as evidence have to make sure the 'clock' has not been reset since the time of death.

Just as the broken watch on a murder victim's wrist in an Agatha Christie novel preserves the time of the crime for posterity, foraminifera preserve temperature

Foraminifera shells are 'reset' by a chemical process called diagenesis. Probably by adding material to the original shell, this tends to shift the ancient temperature signal towards that of surrounding material.

After death the organisms sink to the bottom of the tropical oceans - generally a colder spot than the surface where they lived. The potential of this process to spoil reconstructions of past temperatures has been known for many years.

The only remedy for diagenesis is to find unaffected samples. So researchers have established strict criteria for the selection of reliable shells. Not strict enough, say Paul Pearson of the University of Bristol and his co-workers1, criticizing their own earlier work as much as everybody else's.

Palaeoclimate researchers avoid shells with coarse structural alterations, visible under an ordinary microscope. But in almost all shells from deep ocean sediments, tiny pores, only about a tenth of the diameter of a wool fibre, are recrystallized and altered.

From the sort of specimen now deemed unreliable by Pearson's team, Steve D'Hondt of the University of Rhode Island and Michael Arthur of Penn State University, inferred in 1996 that in the greenhouse world about 66 million years ago, the ocean surface in the tropics was 6 ºC cooler than it is today2.

D'Hondt does not believe that tiny alterations are likely to have made a large difference in his reconstruction. Anyway, better samples are hard to come by, he says: "I don't think there is an ancient foraminifer shell on Earth that has not been altered by diagenesis at some level".

My eyes popped out when I first saw the Tanzanian collection in the Natural History Museum Paul Pearson, University of Bristol

Pearson's group has found some excellently preserved material in Tanzania that should give even more reliable temperature estimates. "My eyes popped out when I first saw the Tanzanian collection in the Natural History Museum," enthuses Pearson, "they look just like modern shells from organisms that died only yesterday." From this glassy-looking material, the researchers estimate that temperatures of the tropical surface oceans 66 million years ago were around 30 ºC, that's 2-3 ºC warmer than today.

Encouraged, Pearson asked geologists worldwide for other pristine samples. This hunt turned up shells of equal quality in Mexico, Alabama and the Adriatic Sea. The material dates back to two distinct greenhouse periods in the Earth's history about 50 and 70 million years ago. During both periods the tropical oceans were warm.

Unfortunately, the locations where these samples were found cast doubt on their reliability. The key to excellent foraminifer preservation is burial in clay-rich material that protects the shells from the surrounding ocean. That such clay-rich sediments occur mainly close to the continents raises the question whether they represent true oceanic conditions. But Pearson's team has taken great care to exclude a significant continental influence on their data. "I am confident that the tropical surface oceans were warm during the Earth's greenhouse periods," says Pearson.

We try to keep models in the realm of what is realistic Lisa C. Sloan, University of California Santa Cruz

The new findings are much easier to reconcile with results from climate models for warm conditions on Earth. Proponents of the 'cool tropic paradox' had suggested that enhanced mixing could explain the cool tropical temperatures. Even so, the models did not reproduce 20 ºC at the tropical ocean surface.

"Of course you can play around with models, but we try to keep them in the realm of what is realistic," explains climate modeller Lisa C. Sloan of the University of California in Santa Cruz. "Our latest results point to tropical greenhouse temperatures 3-4 ºC warmer than present," she confirms.