Effects of changing exposure of thermometers at land stations

In view of the implications for the assessment of climatic changes since the mid-nineteenth century, systematic changes of exposure of thermometers at land stations are reviewed. Particular emphasis is laid on changes of exposure during the late nineteenth and early twentieth century when shelters often differed considerably from the Stevenson screens, and variants thereof, which have been prevalent during the past few decades. It is concluded that little overall bias in land surface air temperature has accumulated since the late nineteenth century: however, the earliest extratropical data may have been biased typically 0.2"C warm in summer and by day, and similarly cold in winter and by night, relative to modern observations. Furthermore, there is likely to have been a warm bias in the tropics in the early twentieth century: this bias, implied by comparisons between Stevenson screens and the tropical sheds then in use, is confirmed by comparisons between coastal land surface air temperatures and nearby marine surface temperatures, and was probably of the order of 0.2"c.