to its mean value as well as with respect to changes in the variance on different time scales. Our findings are mostly mixed, in that the answer depends on the spatial scales selected for analysis.
For the period encompassing the last 100 years, changes in surface air temperature have been greatest and most positive in the period 1970 to 1990. Interannual variability at the largest spatial scales has also increased, although with seasonal differences. By and large, however, increases in variability at these spatial scales are not unique. When one considers smaller regions (in particular different portions of the contiguous United States) the picture changes, so that the 1980s do not necessarily show up as being a unique period in the instrumental climate history of the country.
In order to have a longer perspective on climatic variations, we utilized data from several paleoclimatic reconstructions of growing-season temperature based on tree-ring records and on d18O ratios extracted from different ice-core records. Data from the latter were also taken as a measure of the prevailing air temperature. Annual values were averaged by calendar decades in order to focus on lower-frequency climatic variability. With the exception of the tropical ice cores, the recent decades were not generally unique, in terms of either the average values or increased variability. We note that tropical sea surface temperatures have been generally above the long-term mean since the mid-1970s. It seems plausible that this recent warmth would be reflected in low (less depleted) d18O ratios on the Quelccaya and Dunde ice caps, although more positive d18O ratios were already evident by the 1950s. It should be emphasized again that the exact relationship between actual past climate and these two high-elevation tropical indices is not fully known. Certainly, the period encompassed by the Little Ice Age is reflected in relatively low d18O values, particularly in the Quelccaya record. The development of additional independent ice-core sequences from the Cordillera Blanca in Peru and from extreme western China by L. G. Thompson and colleagues should help in the interpretation of the nature of the climate signals present in these records (e.g., Thompson et al., 1995).
Do we have an answer to the question posed by the title of this study, namely, whether the climate of the twentieth century is different from that of previous centuries? Obviously, the response is not an unequivocal "yes" or "no." As in most instances when one deals with climatic time series, the answer is "It depends." There appear to be differences in regional responses, and they are evident in both the proxy and the instrumental climate records. Insofar as the United States is concerned, the most unusual decade of the last 100 years may have been the 1930s, although the recent period is probably a close second. We can only say that, for most areas of the coterminous United States, the climate of the most recent decades cannot be considered unique, even in the context of the last century.