a role in the Little Ice Age, judging from the pattern of cold, dry winters over Europe (Pfister, 1992). As in the case of altered El Niños, variations in deep-water production could be a natural characteristic of the system, or they might be forced by other mechanisms; the Younger Dryas event may have been initiated by melting of land ice associated with the general deglaciation (Broecker et al., 1988), while the Great Salinity Anomaly could have been a random perturbation on decadal time scales.
How cold was the Little Ice Age, and how much would ocean heat transports have had to be reduced to produce the observed cooling? Mountain glaciers advanced by some 100 to 200 m (Porter, 1975; Broecker and Denton, 1989), equivalent to about one-fifth of their reduction since the full ice age (Rind and Peteet, 1985). If the cooling was also one-fifth as large, we can reduce the North Atlantic sea surface temperatures by one-fifth their full glacial reduction and use a GCM to estimate the resulting regional and global changes in atmospheric temperature. The temperature perturbation associated with this North Atlantic cooling (Table 1), which is reminiscent of a reduction in NADW formation, is shown in Figure 3. Cooling is basically constrained to downstream locations in Eurasia, and immediately upstream in the vicinity of the eastern United States. Note that if sea surface temperatures in other parts of the ocean had been involved (e.g., the North Pacific) the cooling would have been more widespread. Other aspects of the model simulation also appear realistic: Associated with the colder sea surface temperatures there is higher pressure and lower precipitation over Europe, in good agreement with reconstructions for the Little Ice Age (Pfister, 1992).
The ocean heat-transport reduction necessary to produce this cooling is given in Figure 4. Peak reductions are 25 percent in the North Atlantic, concurrent with increased poleward heat transport in the South Atlantic. For the North Atlantic as a whole, poleward heat transports are reduced by 20 percent; this can be contrasted with estimates of North Atlantic transport reduction during the full ice age of some 70 percent (Miller and Russell, 1989). If the heat transport changes are at all proportional to changes in NADW production, it would imply a Little Ice Age change in NADW production two-sevenths the magnitude of the full ice age reduction, a fraction somewhat similar to that of the advance of the ice lines in the Little Ice Age compared with the advance in the full ice age.
The hypothesis of solar forcing of weather and climate is at least a century old (Blanford, 1891). The relationships between sunspots and various weather phenomena have been described in numerous publications, some implying statistical significance. The percentage of variance explained increased dramatically when sunspot-cycle effects were partitioned according to the phase of the equatorial zonal wind, the quasi-biennial oscillation (QBO) (van Loon and Labitzke, 1988). While the reality of this effect is still being debated (e.g., Barnston and Livezey, 1989; Baldwin and Dunkerton, 1989), its existence could have climatic implications to the extent that it would heighten the sensitivity of the troposphere to perturbations (i.e., ultraviolet radiation fluctuations) that directly affect the stratosphere.