The analysis of indirect information about the weather conditions of the past has been particularly useful in establishing the case for shifts in the climate. Often termed ‘proxy data’, these sources occur in many forms and can include almost any form of physical behaviour that reflects the influence of the weather. Ideally, proxy data should record in some permanent form the consequences of seasonal and annual changes of one particular aspect of the weather. In practice, many of even the best records are far more complicated. They often contain information about a variety of meteorological variables. Furthermore, they may be influenced by the weather over a number of years if there is some cumulative effect such as the build-up or decline of groundwater reserves. Finally, there may be problems of disturbance of the records by other external factors, which remove much of the fine detail in the original records, or introduce long-term fluctuations that cannot be easily corrected for or removed from the record.

In this chapter we will concentrate on those records where at least clear annual figures are available, such as tree rings, cores from corals and glaciers (including the ice caps of Greenland and Antarctica), and certain lake sediments. Even these records have some of the drawbacks noted above.