But recent research is questioning the accuracy of the measurements used in many published studies. It now appears tropical trees swell and shrink quite significantly during the day. A behaviour that is not related to actual growth at all.

Foresters use various methods to measure the size of tree stems. One early-developed method is to get an average measurement by wrapping a measuring tape around a set of carefully marked horizontal lines. Though this "Dawkins" method can be very accurate, it is unfortunately very laborious. Two people may be needed and in some cases they have to use ladders on each side of the stem to get up above the buttresses. A second more popular method uses band dendrometers: metal bands that are fixed around the tree, with a scale that indicates any change in diameter. The bands can be left where they are and once in place it is quick and easy to take an apparently precise measurement off the scale. However, these dendrometer band measures have not been critically assessed.

The diameter of a tree stem changes as water is used and replaced. Trees have a reservoir of water in their stems and transpiration demands mean that the tree's vessels carrying the water expand and contract depending on the time of day, weather and water supply. In the tropics a tree with full foliage standing in bright sunshine will use up all the water in its stem. One study in Panama found that about 54 kg of water may be removed from stem storage in a 35-m tall, 1-m diameter tree during the day.

Band dendrometers have, understandably, become the technique of choice for precision growth studies. But a comparison by CIFOR researcher Douglas Sheil in the Malinau Research Forest in Kalimantan, Indonesia, revealed some alarming differences. "The dendrometers are supposed to be accurate to 0.1 mm," said Sheil. "But they only recorded less than a tenth of a millimeter change in just a few stems." In contrast, the Dawkin's method showed all eight stems shrinking by about 1 mm, or 0.1-0.2% of stem size during the day. "We occasionally recorded changes of over 2 mm," said Sheil. The problem is that the dendrometer bands underestimate changes in stem size involving alternating shrinkage and expansion. "Friction probably holds the band fixed while the stems shrink," said Sheil.

Although apparently minor, the errors can influence yield studies. In a forest with an above-ground volume of 400 m3 /ha, then a measuring error of 0.1% represents about 0.8 m3 /ha of tree volume, while a 0.3% bias shows about a 2.4 m3 /ha difference. A survey of the literature confirmed that these measuring differences appeared in published research. Over the years, papers reporting band studies consistently showed small fluctuations of up to 0.1 mm in size, while studies using other methods, found changes up to 10 mm. Numbers like these are large enough to affect the accurate evaluation of woody growth however, Sheil is quick to add that this claim would require further research.

Misinterpretations seem especially likely when the true growth over any measurement interval is small, or when measurements are made under different conditions.

"These patterns of diameter change and associated growth assessment errors are complex and require further study," said Sheil. "Just as long-term sea level changes would be difficult to assess from intermittent records if tides were not recognized, the evaluation of tree growth must consider both long- and short-term fluctuations".