Earlier results have shown that stronger southwesterly winds give higher temperatures, stronger velocities and higher transports in the North Cape Current (NCaC) in the western Barents Sea. This investigation shows that the wind field causing the stronger southwesterlies also causes a wider NCaC, i.e., the NCaC becomes warmer and wider at the same time. In warm periods, the NCaC seems to be a two-core current system, while it has only one wider core in cold periods. The reason for the fluctuations in the width of the NCaC is a non-uniform wind field across the current. The varying width of the NCaC affects the location of the Polar Front south of Bear Island. The location of this front is not as stationary as earlier believed and in warm periods with strong wind, the front is located further upslope than in cold periods.
 The part of the Norwegian Atlantic Current (NwAC) that enters the Barents Sea is called the North Cape Current (NCaC). In the western Barents Sea it splits in two main branches; one branch continues eastwards parallel to the coast while the other turns north along the Hopen Trench (Figure 1). Some of the Atlantic Water (AW) recirculates in the Hopen Trench and leaves the Barents Sea as a westward current topographically trapped on the slope to the south of Bear Island. Note that during winter, there may be a northeastward flow of AW on parts of the slope [Johansen et al., 1988; McClimans and Nilsen, 1993; Li and McClimans, 1998]. North of the AW, the Polar Front separates this water mass from the water of Arctic origin flowing westwards in the Bear Island Current. The position of the Polar Front is determined by the barotropic circulation of AW and is strongly tied to topography [e.g., Johannessen and Foster, 1978; Harris et al., 1998]. South of the NCaC, the Norwegian Coastal Current transports fresh water masses into the Barents Sea.
 The variability of the Atlantic flow has a strong coupling to the regional wind field. Stronger winds give higher transports and a warmer current both in the Norwegian and the Barents Seas [Ådlandsvik and Loeng, 1991; Loeng et al., 1997; Blindheim et al., 2000; Orvik et al., 2001]. In the Norwegian Sea the stronger winds also give a narrower current [Blindheim et al., 2000]. However, what happens with the lateral extent of the NCaC in the Barents Sea is not known. Recent investigations indicate that the meridonal position of the water entering the Barents Sea to a large extent determines where it ends up within the Barents Sea [Asplin et al., 2005]. A varying width of the NCaC may therefore have large impacts on the local climate. In this study the lateral extent of the NCaC is investigated, and the consequence for the movement of the Polar Front is evaluated.
2. Data and Analysis
 Hydrographic data have been sampled six times a year since 1977 in a section between Norway and Bear Island (Figure 1). Time series of temperature and salinity anomalies in the AW core were calculated in the part of the section where the main Atlantic inflow takes place as defined by salinity (Figure 2). The seasonal signal was removed from the data prior to the calculations. A time series of the cross-sectional area of AW (T ≥ 3°C and S ≥ 35) was calculated for the entire section. To investigate the details of the variability, the area was also calculated separately in two boxes in the northern and southern parts of the section (Figure 2). The seasonal signal was not removed prior to the area calculations because the total area does not show any pronounced seasonal cycle (Figure 3d). The time series was compared to the North Atlantic Oscillation (NAO)-index updated from Hurrell . Finally, an empirical orthogonal function (EOF) decomposition was performed on the temperature data to identify the dominant mode of variability in all the temperature data (not only AW).
 The time series (Figures 3a–3c) revealed the already known result that for the period studied here both the temperature and the salinity in the AW core fluctuate in phase with the NAO winter index [e.g., Dickson et al., 2000]. However, there is a lag of about 1 year between the 3-years running means of NAO and of temperature, and about an additional year to the 3-years running mean of salinity. The correlation between the 3-years running means of the NAO winter index and of January temperature anomaly delayed 1 year is 0.76. What is more surprising is that the total area of AW in the section also varies in phase with these parameters (Figure 3d), with only a 2–6 months lag compared to the temperature. The correlation between the 3-years running means of the temperature anomaly and of AW area delayed 6 months is 0.92.
 When investigating the amount of AW in the northern and southern ends of the section some differences appear (Figures 3e and 3f). The area in the northern end fluctuates strongly in phase with the total area and the temperature anomaly in the Atlantic core. The correlation between the 3-years running means of temperature anomaly and of area in the northern end is 0.88, and the delay is only 2–3 months. This means that the temperature in the Atlantic core actually is a very good indicator of how much AW there is on the slope south of Bear Island. On the other hand, the AW area in the southern end does not show a good correlation with the temperature anomaly (the correlation between the 3-years running means is only 0.31). In fact, for the area of AW in the southern parts, the seasonal variability is dominating, while for the total area and the area in the northern parts the interannual variability is much stronger than the seasonal.
 The southern limit of the northern box (73°45′N) corresponds roughly to the 280 m depth contour on the slope south of Bear Island (Figure 2). According to Gawarkiewicz and Plueddemann  and Harris et al.  the location of the Polar Front is fixed at the 260 m isobath by the barotropic circulation of AW. If so, the area in the northern box should have a relatively stable value close to zero all the time. The fact that this area varies in phase with the temperature in the Atlantic core and the NAO index, indicates that the location of the Polar Front is moved upslope during warm and windy conditions. This is illustrated by plotting different water masses in the section for a cold and a warm year (Figure 4). In cold years the front may be wide and AW may not reach the slope at all, while in the warm year of 1983 the front is very sharp and AW is found all the way up to about 75 m depth. The difference in the northward extent of AW is 130 km in these two years. Although the details of this depend on how the water masses are defined, it shows that even the base of the Polar Front is clearly moved upslope during warm conditions.
 The positive correlation between the temperature anomalies and the area, and the upslope motion of the Polar Front, may be caused by a warmer and wider NCaC at the same time. However, it is also possible that the correlation is due to a stronger, or wider, or both stronger and wider flow in the westward AW current on the southern slope of Bear Island. To identify the actual reason, an EOF analysis was performed on the data set. This has the advantage that the analysis includes all the temperature data at the same time. The leading EOF shows a positive sign at the entire section (Figure 5), i.e., a simultaneous temperature increase (or decrease) in the coastal current, the NCaC and the westward-flowing Bear Island Current and bottom currents. The only thing that can cause this is a simultaneous northward movement of the Polar Front and southward movement of the coastal front. If the fluctuating area and the cross-slope motion of the Polar Front is due to only a stronger westward flow of AW, there should be a lag between the temperature increase in the inflow area and the outflow area. The correlation between the 3-years running means of the time series of EOF1 and the temperature anomaly in the AW core is 0.99, i.e., the NCaC gets both warmer and wider with the stronger winds. The results contrasts the situation found in the Norwegian Sea where stronger southwesterlies are associated with higher temperatures but a narrower current [Blindheim et al., 2000; Mork and Blindheim, 2003]. The different response in the two areas is connected to the wind field, and to the different structure of the currents (the NwAC consists of an inner barotropic and an outer baroclinic branch, while the NCaC is mainly a barotropic branch). In the Norwegian Sea a positive NAO index gives, in general, strong southwesterlies across most of the NwAC (Figure 6), pushing it towards the Norwegian continental shelf slope thereby giving a narrower current [Blindheim et al., 2000]. In addition, a positive NAO index may strengthen the inner barotropic branch of the NwAC and weaken the outer baroclinic branch [e.g., Mork and Blindheim, 2000]. The narrower current reduces the heat loss to the atmosphere, thereby increasing the temperature. At the same time the stronger winds increase the transports [Orvik et al., 2001], which also contribute to the temperature increase. In the western Barents Sea, the positive NAO index gives stronger southwesterlies in the south (Figure 6), pushing the NCaC towards the Norwegian coast. At the same time there often is stronger northeasterlies in the north, pushing the AW and the Polar Front northwards (Figure 6). Because the stronger wind also adds momentum [Ingvaldsen et al., 2004], the current may be wider without losing speed. Because the temperature of the inflowing water is mainly determined by the temperature upstream in the Norwegian Sea (and that temperature is increasing due to the stronger winds there), the temperature in the western Barents Sea will also increase. This means that it is possible to increase both the temperature and the width of the NCaC at the same time.
 A stronger inflow of AW may increase the westward outflow on the slope south of Bear Island 1–2 years later (when the water has passed around the Hopen Trench and if the increased inflow has not been channeled into the eastern Barents Sea). This means that in addition to a wider NCaC, the observed changes in the 3-year running means may be due to a stronger, wider, or both stronger and wider westward current on the southern slope of the Bear Island. Unfortunately, it is not possible to determine this from hydrography data alone. To investigate this, the direction of the currents (from direct measurements or models) must be examined for both warm and cold periods.
 Another interesting aspect with the EOF analysis is that it shows that the variability is strongest in two cores, one core between 71°N and 71°30′N, the other core centered near 72°30′N (Figure 5). Earlier investigations have shown that the NCaC may be a two-core system with the cores near the locations above, or as one wider core [e.g., Haugan, 1999; Ingvaldsen et al., 2004]. This analysis indicates that the splitting in two cores is most pronounced in warm and windy conditions. This aspect is impossible to investigate in detail because the time series of the available current fields does not span both cold and warm periods. Surrogate current fields may be simulated in numerical models, but that is beyond the scope of this investigation.
 Earlier results have shown that stronger southwesterly winds give higher temperatures, stronger velocities and higher transports in the western Barents Sea [Ådlandsvik and Loeng, 1991; Loeng et al., 1997; Ingvaldsen et al., 2004]. This investigation shows that the wind field causing the stronger southwesterlies also causes a wider NCaC, i.e., the NCaC becomes warmer and wider at the same time. This contrasts the situation in the Norwegian Sea where the NwAC gets warmer and narrower with stronger southwesterlies [Blindheim et al., 2000; Mork and Blindheim, 2003]. The results also indicate that in the warm periods, the NCaC is a two-core current system, while it has only one wider core in cold periods. The reason for the fluctuations in the width of the NCaC, is the generally non-uniform wind field across the current.
 The results show that the location of the Polar Front south of Bear Island is not as stationary as earlier believed. The location of the front varies in phase with the climate of the Barents Sea. In warm periods with strong winds, the front is located further upslope than in cold periods.
 This work was funded by the European Union project ASOF-N and the Research Council of Norway through the project “Variation in space and time of cod and other gadoids: The effect of climate and density dependence on population dynamics.” I appreciate the comments of the reviewers, who contributed most usefully to the final revision of the paper.