Theophrastus, a friend of Aristotle and pupil of Plato, first noted that 'the ends and beginnings of the lunar month are apt to be stormy', while Francis Bacon recorded that a 35-year rhythm was the subject of inquiry in the Low Countries at the beginning of the 17th century.
Subsequent analysis has come up with a more complicated picture. From daily fluctuations to the massive changes associated with the ice ages, the climate is constantly changing.
Most interest has focused on whether there are cycles on a time-scale that corresponds to the human lifespan. If they exist, and we knew what caused them, we would be able to forecast, years in advance, the good and bad summers and longer periods of drought. Sadly, although records show evidence of periodic behaviour, the cycles come and go. Moreover, in different parts of the world the records show examples of cycles that last anything between one and 100 years. The net effect is a tantalising glimpse of possible order.
The best example of a weather cycle is the quasi-biennial oscillation in the stratospheric winds high above equatorial latitudes. The direction in which these blow reverses itself approximately every two years - a pattern that has been studied since the 1950s. But there is no proof that it is linked to its much weaker cousin, often detected in ground-level weather records.
Faint cycles that last between three and five years are believed to be related to fluctuations in the temperature of the equatorial Pacific. Known as the El Nino Southern Oscillation, it affects the weather throughout the tropics and subtropics. A recent paper in Nature showed that the maize crop in Zimbabwe is closely linked to the behaviour of the El Nino. But although this suggests the possibility of forecasts that stretch months ahead, it falls a long way short of a cycle that can be predicted years in advance.
Many putative cycles of 10 years or longer have been linked with changes in the number of sunspots. Ever since Heinrich Schwabe published his observations in 1843, researchers have been seeking evidence of a similar 11-year cycle. This has been extended to a 22-year period to reflect the fact that the magnetic polarity of pairs of sunspots reverses with every cycle. Also, the number of spots shows longer-term fluctuation of around 80 to 90 years, as does the length of the sunspot cycle which varies about the 11-year mean.
The search for an 11-year cycle has not produced a convincing case for changes in sunspot numbers influencing the weather. The 22-year cycle is more difficult to deal with. There is a ubiquitous, if weak, 20-year cycle in many weather records, including measurements of global temperature. This may be linked to sunspots, but there is a continuing debate as to whether an 18-year variation in the lunar tides is a better explanation for the observed fluctuations.
There remains one awkward fact: the correlation between changes in global temperature and both sunspot numbers and the length of sunspot cycles since 1880. Recent satellite measurements show that the variation of the output of the sun does vary by the order of about 0.1 per cent during the sunspot cycle. This is, however, far too small to explain observed changes in global temperature. This search for order may be misguided. Results from cores recently taken from the Greenland ice-cap show that the climate during the past 10,000 years has been stable. Prior to that, there were changes on almost every time-scale. Even more intriguing is evidence that during the previous warm interglacial period the climate behaved equally erratically.
This picture of chaos is in stark contrast to our view of a relatively orderly climate. It underlines the fact that not only the comparatively modest changes in recent centuries, but also much greater sudden shifts in the distant past, may be nothing more than evidence of the ablity of the climate to shift dramatically of its own accord.
The conclusion makes gloomy reading for those looking to find order in the weather. It also has a more worrying implication. Changes brought about by human activities could jolt the climate out of the relative stability of the past 10,000 years. Instead of worrying about a rise in global temperatures, we could be into vast and sudden changes, and weather cycles will be the least of our problems.
The author's book 'Weather Cycles: Real or Imaginary?' will be published in the autumn by Cambridge University Press at pounds 14.95.