during the 1740s, 1780s, 1800s, and 1810s, and the late 1830s and 1840s. Warm decades are limited to the 1820s. Similar decadal similarities are seen in other regions. The two Asian records show cool decades during the 1830s and 1840s as in Europe. The four North American records show relative cold during the 1780s, 1810s, and 1830s, again similar to the European data, although warmth is indicated during the 1800s, the 1820s, and to some extent during the 1840s. The cold-warm-cold oscillatory pattern between the 1810s and 1830s is particularly striking in the New Haven and Toronto records in North America and at Stockholm and St. Petersburg in Europe.
Since 1850 the best-known difference between European temperatures and those in the rest of the hemisphere occurred during the early 1940s. At this time Europe was cold while the rest of the hemisphere was warm, at a level exceeded in magnitude only during the 1980s (Jones et al., 1986a). North American station records show warmth during the 1800s and 1840s, while European stations show cool conditions. Although some correspondence appears between warm and cool decades over the northern landmasses, more comprehensive regional studies of twentieth-century temperature variations over the Northern Hemisphere have shown that no one region can be said to be fully representative of hemispheric-wide temperatures (Briffa and Jones, 1993; Jones and Kelly, 1983). There is, therefore, no reliable short-cut method of estimating hemispheric or global temperatures on the basis of relatively few station records, at least on the decadal time scale, as has been tried by Groveman and Landsberg (1979). (Lack of longer-time-scale data prevents reaching a similar conclusion about estimating temperatures on century time scales.)
Many methods are used to reconstruct past climates (see Bradley, 1985). Only a few, however, have the potential for providing records that can be resolved to an annual or seasonal level. Of those that can, the most widely used sources are documentary historical data, dendroclimatic data, and ice-core records (see Bradley and Jones, 1992). Documentary reconstructions are mainly restricted to Europe and Asia and to the period since about A.D. 1400 (see, e.g., Mikami, 1992). Few documentary reconstructions extend continuously over long time periods, the proxy temperature series for Switzerland from 1525 (Pfister, 1992), for China from 1470 (Wang and Wang, 1991), and for Iceland from 1590 (Ogilvie, 1992) being the longest. Documentary evidence probably represents the most reliable source of past climatic information on short time scales. The ability of some documentary indicators to faithfully record low-frequency climatic variations on time scales longer than the human life span (in particular a century or more) must be questioned, however.
Ice-core evidence is limited to high-latitude and high-altitude locations. Furthermore, few ice-core reconstructions of temperature are explicitly calibrated against instrumental time series, and there may be some ambiguity in interpreting the seasonality of the climate variations (Bradley and Eischeid, 1985; Thompson et al., 1986). When isotope-temperature relationships have been tested, the best correlations on the interannual time scale explain only about 25 percent of the temperature variance (Jones et al., 1993; Peel, 1992). Notwithstanding these points, ice cores represent important environmental information, and may be the principal means of reconstructing highly resolved long temperature series in terrestrial tropical areas (albeit at high elevations).