the north the hemispheric mode is dominat- ed by the North Atlantic and Arctic regions, with only a weak link to the north Pacific (C. Deser, Natl Center for Atmospheric Res., Boulder, Colorado) 4 . Which of these views of the NAO is the more useful therefore depends on the research agenda. For instance, for studies of seasonal predictabili- ty in the Atlantic region it is usually better to concentrate on the NAO. Another aspect is that the NAO may not always have shown the same behaviour. Indi- rect data for the past four centuries suggest that the amplitude of variability might have increased significantly after the end of the Little Ice Age, around 1850 (E. Cook, Lam- ont–Doherty Earth Obs., Palisades, New York). Moreover, the consequences of the pattern, for example on sea-ice extent in the Arctic (J. E. Walsh, Univ. Illinois, Urbana- Champaign), depend on the exact positions of the two pressure centres — which can dif- fer significantly in years with the same index value. The persistent positive trend in the index since about 1960 raises the question of whether global warming is responsible. There is no clear answer. A new index, based on surface pressures from Gibraltar, Reyk- javik in Iceland and Ponte Delgada in the Azores, shows a similar trend for the early part of the twentieth century (P. Yiou, Lab. des Sciences du Climat et de L’Environ- nement, Gif-sur-Yvette), implying that nat- ural variability can account for Europe’s mild and wet winters since 1960. But an analysis of the Lisbon–Iceland index indi- cates that natural variability can be rejected as a cause for the recent trend (S. B. Feldstein, Univ. Pennsylvania). After considering these conflicting results and the processes that drive variations of the NAO, most partici- pants, asked to give their best guess for the coming decade, expected the index to increase further, rather than decline towards the mean as would be expected in the absence of a link with climate change. Two main candidate mechanisms emerged that might explain the long-term persistence of trends in the lower atmos- phere (in which disturbances can normally be tracked for only about a week). Interac- tions with the ocean are one possibility. A statistical analysis of observations reveals two patterns of Atlantic sea surface tempera- tures that tend to precede specific phases of the NAO: positive anomalies in mid-latitude temperatures induce a positive phase with a lead time of six months, and the tropical Atlantic surface temperatures exert a weaker influence on timescales of up to two months (A. Czaja, Massachusetts Inst. Technol.). The other candidate mechanism is atmospheric, and centres on the polar stratospheric vortex — a rotating circulation pattern in the higher levels of the atmosphere above the poles. Depending on its strength, the vortex acts as a window that allows atmospheric waves originating in the tropo- sphere to pass into the upper atmosphere, or as a mirror reflecting the waves back down. This second phenomenon could amplify a positive NAO phase if there is interference between the reflected waves and the original waves (H.-F. Graf, MPI für Meteorologie, Hamburg) 5 . Both sea surface temperatures and the strength of the polar stratospheric vortex are thought to increase with increasing green- house gases in the atmosphere. So this con- sideration points to a causal connection between global warming and the positive trend of recent years. If this is indeed the case, there could be a beneficial effect: simulations with a general circulation model suggest that a positive NAO may temporarily mitigate one of the most worrying possible impacts of global warming, the slowing of the thermo- haline circulation in the North Atlantic (T. L. Delworth, Geophys. Fluid Dynamics Lab., Princeton Univ.). This circulation is one of the main engines of heat transport around the globe — if it slowed or stopped, the cli- matic consequences would be severe. The NAO has a considerable impact on natural ecology and human economies, in arenas ranging from plant and animal popu- lation dynamics to energy production in Scandinavia. So many researchers are attempting to predict its behaviour on both seasonal and decadal timescales. At the sea- sonal scale, the aim is to exploit the predictive value of sea surface temperatures. According to one forecast, in which observed Atlantic sea surface temperatures for May and November 2000 were fed into a model, there is a 66% chance of a positive NAO for the coming winter (M. Rodwell, Hadley Centre, Reading, Berks.). If increasing greenhouse- gas concentrations in the atmosphere are driving the current positive phase, we should expect further increases in the frequency of mild, wet winters in northern Europe. But to confirm whether global warming does indeed change the balance between the posi- tive and negative phases, we will simply have to wait until a clear, tell-tale signal emerges from the noise of natural variability. ■ Heike Langenberg is an associate editor at Nature. 1. Hurrell, J. W. Science 269, 676–679 (1995). 2. Thompson, D. W. J. & Wallace, J. M. Geophys. Res. Lett. 25, 1297–1300 (1998). 3. Walker, G. T. & Bliss, E. W. Mem. R. Meteorol. Soc. 4, 53–84 (1932). 4. Deser, C. Geophys. Res. Lett. 27, 779–782 (2000). 5. Perlwitz, J. & Graf, H.-J. J. Clim. 8, 2281–2295 (1995). news and views NATURE | VOL 408 | 21/28 DECEMBER 2000 | www.nature.com 925 Figure 1 Deviations from a 52-year average of sea-level atmospheric pressures over the North Atlantic region, during winters with a high North Atlantic Oscillation index (1 standard deviation above the average). Anomalously large differences between pressures over Iceland and Lisbon lead to a strong storm track in the North Atlantic, bringing mild, wet and stormy weather to northern Europe. The data are derived from measurements for the months December to March, for 1948–99. Units are hectopascals (ǃmillibars). (Graphic courtesy of Thomas Jung and Michael Hilmer.) –1 –1 1 1 2 0 0 0 0 0 –2 –3 T he development of materials with dimensions in the nanoscale range is a highly active area of research because they often have very different properties from the bulk material. These properties can offer new or improved technological appli- cations. So far, materials scientists interested in nanotechnology have mostly focused on semiconductors and ceramics. For example, semiconductors made from nanocrystals have unusual optical, electrical and magnet- ic properties, and ceramics made from nano- particles have greater hardness and plasticity than normal. However, on page 946 of this issue, Maier and colleagues 1 demonstrate that restructuring simple ionic crystals at the nanoscale can also alter their electrical properties, pointing the way to potentially Nanotechnology Solid progress in ion conduction Alan V. Chadwick Materials that conduct ions are useful in devices involving electrochemical reactions, such as fuel cells and batteries. Low ionic conductivity was a problem for these materials until researchers built nanoscale versions. © 2000 Macmillan Magazines Ltd