By Jonathan Amos
BBC News science reporter, San Francisco
The close link between temperatures in the North Atlantic and the strength of ocean circulation is underlined by a new analysis of sea-floor sediments.
Foam marks points where samples have been extracted from the core. (Image: Tom Kleindinst, Woods Hole Oceanographic Institution)
The sediments were drilled from Blake Outer Ridge off the US east coast.
They contain traces of naturally occurring radioactive atoms in ratios that are a giveaway for the speed of ocean waters going back 60,000 years.
The work by a team from the Woods Hole Oceanographic Institution is critical to our understanding of climate change.
If human activity alters circulation speeds, as many scientists suspect may happen, it could have a dramatic impact on temperatures.
This is of particular concern in northern Europe, which benefits from the strong flow of warm waters that sweep past it from the tropics and keep winters mild.
These waters eventually sink, or overturn, at high latitudes and return to the tropics as a deep, cold flow.
If this great circulation, or conveyor, slows - and there is some evidence this is happening - then European winters should become harsher.
"Warm periods in the past are generally associated with strong ocean circulation, or overturning; and cold periods are generally associated with a weak overturning circulation," said Dr Candace Major, from Woods Hole.
"We have documented these changes in the past and associated them with abrupt climate change."
Dr Major was speaking here at the American Geophysical Union (AGU) Fall Meeting.
She was presenting research by her team that builds on work reported last year by another group. The 2004 results linked past temperatures, worked out from Greenland ice cores, to the strength of ocean circulation for the past 20,000 years.
The new data greatly extends that record and reinforces the significance of the relationship.
It is done by using what scientists call a geochemical proxy - an indirect record of past activity.
In the case of ocean sediments, this is a ratio of two atoms - protactinium and thorium - that appear in the water as naturally occurring uranium undergoes radioactive decay.
It happens that as sediments fall through the water, thorium tends to "stick" to them and is buried; while the not-so-sticky protactinium has a tendency to be washed out of the North Atlantic basin.
"So, the stronger the circulation, the more the protactinium will be exported out of the North Atlantic and the less of it will be buried in the sediment," explained Dr Major.
"In the case of a sluggishly moving ocean, the more will be buried in the sediments."
Dr Major displayed graphs here tracking temperature and ocean circulation from about 25,000 to 60,000 years ago that were beautifully aligned - they were virtually one for one.
The big question for science now is, which way will the graphs go in the future?
Modelling indicates global warming will slow North Atlantic circulation.
What happens is that as Arctic ice melts and Arctic rivers flow faster - trends which have both been documented - the northern ocean fills with fresh water and becomes less salty.
Less salinity means a lower density; the waters then cannot overturn, so the circulation weakens.
The previous modeling work had predicted that if this "great conveyor" turned off completely, Europe would cool by perhaps four to six degrees Celsius.
And Andreas Schmittner, of Oregon State University, came to the AGU to present the results of the very latest work in this field - a broad analysis that combined nine different computer models of future climate change.
These suggested there would not be a shutdown in the next 100 years, but rather a slowdown of about 25%.
This produced a fair amount of muttering among journalists in the hall who just this past fortnight had reported on actual measurements collected in the North Atlantic which showed there had already been a 30% weakening in strength in the past 50 years.
"This is an apparent contradiction between the models and the observations," conceded Dr Schmittner. "We have to reconcile these differences, obviously. We have to get more observational data and improve our models."
Dr Major Commented: "It could be that the models are pointing - if you like - to average trends in ocean circulation, but that there could be significant inter-annual variability which is what the observational data has picked up; we just don't know."