The last part of Eloise Gibson's series talking to climate experts about what they don't know, what they wish they knew and how they can find out more.
Scientists are chipping away at a glitch in the climate records, hoping to explain why tree-rings that track temperature changes successfully until the 1950s suddenly veer off.
Researchers believe global warming or other man-made changes may be to blame for an unexplained slowdown in growth of some of the ancient trees used to track temperatures back more than 1000 years.
Although tree-rings records appear to be a good measure of temperature changes most of the time, nobody knows why the usual techniques used with them fail from the 1960s.
The divergence between tree-rings and real, thermometer measurements in the past 50 years has been used to attack temperature reconstructions used by the International Panel on Climate Change and others - particularly since emails from the University of East Anglia's climatic research unit revealed unit head Phil Jones discussing "hiding the decline" in tree-ring temperatures.
It turned out the email referred to a common technique of replacing tree-ring records with direct thermometer measurements from 1961, when tree-rings show temperature declining while real measurements from thermometers do not.
But although researchers of ancient climate have several theories, so far no single theory can explain why some Northern Hemisphere trees behaved differently in the past few decades.
Dr Andy Reisinger, a climate researcher at Victoria University who has followed the progress of proxy temperature reconstructions, said it could be that a lack of rain in recent decades had stunted tree growth in some high-altitude spots - or that when temperatures reached a certain point, trees began to react differently.
Whatever the cause, "the relationships [between tree-rings and temperature] that we've developed for the last 500-100 years may not apply in the last 50," he said.
So far, Dr Reisinger said tree-ring records from the Southern Hemisphere were limited.
"Tree-rings always only tell you something about a specific location and therefore tree-rings are always selective," he said.
Tree-ring records study the width of tree-rings - the wider the ring, the more the tree grew that year.
Generally trees grow more in warmer years.
Tree-ring records are often combined with other ancient reconstructions to form a "hockey stick" pattern, which shows late 20th century temperatures rising sharply from the long-term average.
Those reconstructions helped the IPCC conclude that the last 50 years of the 20th century were probably the warmest in the Northern Hemisphere in more than 1000 years.
Dr Reisinger said that for most of the record tree rings match other physical evidence from ice cores, sediment records, stalagmite and coral fairly closely.
There is no way of confirming the picture with real thermometer temperatures until 1850, although tree-rings are fairly accurate for most of the period when thermometers overlap.
Meanwhile, New Zealand researchers at Auckland University and the National Institute of Water and Atmospheric Research, including tree-ring specialist Andrew Lorrey, may be able to add to the picture by building a long-term climate record from Kauri stumps many thousands of years old.
Massive kauri trees found buried in Northland bogs have yielded some of the world's oldest tree-ring records - some about 30,000 to 60,000 years old.
Some kauri rings are already being used to build a partial record of droughts.