Abstract
Although climate should not be reduced to temperature, it is the simplest parameter to describe and quantify. Global warming over the lasttwo hundred years, and particularly since 1970, is undeniable, but some see it as the result of urban development rather than atmospheric disturbance. Today, temperatures are measured both in situ and by satellite. To estimate temperatures of the more distant past, indirect observations are necessary. As observations are very heterogeneous, how can we deduce an average temperature ? Do we see an increase in the frequency of extremes, or simply a shift in the mean and distributions ?