By Paul Rincon
BBC News science reporter
A row has erupted over one of the most provocative symbols of global warming - the iconic "hockey stick" graph.
This record of Northern Hemisphere temperature variation in the last 1,000 years shows a recent warming trend apparently linked to human activities.
New work in the journal Geophysical Research Letters questions its validity, challenging the way it was originally put together.
Other scientists say the work simply highlights a technical issue.
These scientists point to the fact that the recent and dramatic warming trend seen in the chart has been replicated by other teams using different raw data and methodologies.
The hockey stick graph originally appeared in a research paper published in Nature in 1998 by US scientists Michael Mann, Raymond Bradley and Malcolm Hughes. It shows a line that is roughly straight for 900 years and then a sudden upturn in temperatures after 1900.
The authors of the 1998 paper drew on a variety of "proxy" sources that give information on past climate. These included tree rings, ice cores, coral, instrumental data and historical records.
The simple and dramatic way the graph illustrates recent global warming - not to mention its inclusion in a key report by the United Nations' Intergovernmental Panel on Climate Change (IPCC) - has made the hockey stick a talisman of sorts for those advocating action to curb the release of greenhouse gases.
It has also become a target for researchers who dispute global warming is a result of present human activities.
The new controversy surrounds a mathematical technique used to construct the hockey stick graph.
Some question whether human activities really cause warming
At one stage in their original calculations, claims the latest study, Mann and colleagues applied a particular statistical convention to a data set consisting of tree rings from North America.
This convention favoured a hockey stick shape for the final temperature record, say Stephen McIntyre and Ross McKitrick in Geophysical Research Letters.
Consequently, claims Dr McKitrick, of the University of Guelph in Ontario, Canada, the inclusion of this data set could "flip" the entire analysis.
"Our work is very significant for understanding the way that particular study got the conclusions it did," Dr McKitrick explained.
"What we've been able to show is that their method looks through the data for hockey stick shapes and then promotes those to the dominant pattern," he told the BBC News website.
The authors say that when they took this into account, the results in the 1998 paper did not achieve statistical significance.
"This means their model is not qualified for projecting temperatures back more than 500 years," Dr McKitrick added.
However, supporters of a recent warming trend claim that even when using another convention, the hockey stick shape stays the same.
"This is a tiny step in the hockey stick analysis. If you do it in different ways, you still get the answer you got before, providing you don't throw away any significant data," said Gavin Schmidt, of the US space agency's (Nasa) Goddard Institute for Space Studies in New York, US, who has worked in the past with Michael Mann.
Dr Schmidt points out that McIntyre and McKitrick use a different convention but do not alter subsequent steps in their analysis to account for this.
As a result, he says, McIntyre and McKitrick's analysis removes crucial data included in the original hockey stick work.
The authors of the present study have published previous critiques of the hockey stick work.
"They keep going on about one data set, but there are loads of others that show the same thing," Professor Phil Jones, director of the Climate Research Unit (CRU) at the University of East Anglia, UK, commented.
He points to a paper published in February in Nature by Anders Moberg and colleagues that found the late 20th Century was the warmest period in at least the past 2,000 years, using data from climate proxies such as lake sediments and stalagmites.
However, even those scientists who agree with evidence for recent, strong anthropogenic warming have reservations about models such as the hockey stick.
According to Professor John Waterhouse, of Anglia Polytechnic University in Cambridge, UK, the so-called Medieval Warm Period (AD 800 to 1400) and the Little Ice Age (AD 1600-1850) do not show up.
"Most climate researchers expected them to be there," he told the BBC News website.
He added that uncertainties in the methods used for climatic reconstruction underestimate the changeability of climate in the past.