The discovery began, when researcher Georges Mabille of the climatology and topo-climatology laboratory, which is part of the Geography Department, decided to closely study the total hourly readings taken since 1966, at the Bierset meteorological station, located a few kilometres away from the town of Liège. His objective was to identify possible cycles. There was a phenomenal quantity of data involved. To help him in his analysis, the climatologist solicited the help of Samuel Nicolay of the Mathematical Institute of the University of Liège.

A preliminary analysis, relying on the traditional Fast Continuous Fourier Transforms method of calculation, made it possible to focus only on periods of 24 hours and one year. These periods showed temperature ranges of 2 and 7 degrees respectively. This was immaterial, as, rather than taking into account the absence of other periodic results, Samuel Nicolay changed his approach by applying a second, more powerful, mathematical tool, which is even more complex than the Fourier Transforms, known as wavelets. He had already used this mathematical analysis in his study of DNA sequencing, in order to reveal the asymmetries inherent therein. (Read the article DNA : Next Target)

When applied to climatic data, the wavelets have also thrown up a lot of surprises. The twenty-four hour and yearly cycles clearly re-emerge as dominant periods. But this time they are accompanied by a weak cycle of 30 months, characterised by a temperature range of close to half a degree.

It is difficult in principle to connect this observation with any physical mechanism. «In order to verify this result, we began by applying the same analysis to data coming from another meteorological station, in this case, the station at Uccles, which has a record of daily temperature measurements since the 1900’s. A period of 30 months emerges from the measurements. Finally, it was to be found in the data gathered by more than several dozen European stations.» explains Samuel Nicolay.