Since the characteristic time scale over which atmospheric circulation anomalies develop is on the order of 8 days, the atmosphere, with it's short-term memory, provides no means of driving the longer-term variability of the NAO (Frankignoul, 1983; Hoskins, 1983) . The ocean, however, with a large heat capacity and long-term memory, may set the pace for observed interdecadal variations. Testing the original hypothesis of Bjerknes (1964), which links variations in North Atlantic SST on the interdecadal time scale to swings in the intensity of the Gulf Stream, may provide a means of understanding the oceanic component of the NAO. Once this oceanic component is more clearly understood it may be possible to forecast intense changes in the NAO.

Although the atmospheric component of the NAO has been visible since the 1700's, continued study of an albeit short hydrographic data record is essential to understanding the oceanic component and what role it may play in modulating the long-term NAO signal. McCartney and Curry (1996) have proposed an oceanic oscillator, building upon the work of Bjerknes (1964). Their oscillator is a "pipeline" of water fed by the Gulf Stream which circulates around the subpolar gyre on a timescale of roughly 20 years, comparable to the NAO's long-term period. A gradual warming of the pipeline in the 1950's and 60's, is seen to correspond with weaker westerlies across the Atlantic basin and a negative NAO. The pipeline, for reasons that are still unresolved, began to run cold in the 80's leading to a strengthening of the westerlies and a more positive NAO.

Determining whether these fluctuations are the signatures of self-sustained deep ocean circulation (Latif and Barnett, 1996; Weaver and Sarachik, 1991) or part of a dynamic ocean-atmosphere interaction is the focus of current efforts. Modeling studies (Battisti et al., 1995; Delworth, 1996) suggest that thermohaline circulation may behave as a damped oscillatory system at decadal to interdecadal frequencies, where SST variability is primarily driven by atmospheric perturbations to the surface latent and sensible heat flux. The results of these modeling studies suggest that the NAO is an atmospheric mode of instability, driven by land-sea temperature and moisture contrasts and independent of feedbacks (i.e., it can operate with a fixed SST boundary layer). The instability mechanism, however, is quite complex, and further begs the question of whether coupling between the ocean-atmosphere is necessary to explain the oscillation itself.