in the southwest quadrant of tlie eddy and 

 tracked by LOR AN C for 12 hours (fip. 27). 

 The drotrue was tlien recovered and re-deployed 

 in the eddy's northern quadrant and tracked for 

 about 36 hours (fig. 28). The tracks of the 

 drojrue can be accounted for satisfactorily by 

 assuming that prior to and during the drogue 

 experiment the eddy drifted soutliward at a 

 speed of about 0.13 knots, and that the current 

 acting on the drogue was the vector sum of the 

 geostrophic flow in the eddy and a simple wind 

 driven current as described in the National 

 Search and Rescue Manual {l!>7;i). The esti- 

 mated average winds for the tracking episodes 

 are shown in Table 3. The elTect of inertial 

 currents can be seen in botli of the drogue tracks. 

 During the end of the eddy survey a storm was 

 in progress with winds from the northeast quad- 

 rant of the compass at "20 to 25 knots. At about 

 1600Z on 11 August the wind dropped to lo 

 knots. This would have permitted an inertial 

 current to begin rotating. The inertial period 

 at the latitude of the eddy is 19.5 hours. It 

 appears from figure 27 that the majority of the 

 12 hour drift of the southwest quadrant drogue 

 track occurred predominantly during the por- 

 tions of the inertial period in which there was a 

 northward component to the inertial current. 

 This would account for the northward displace- 

 ment of the drogue after 12 hours relative to the 

 position indicated by the combination of wind 

 and geostrophic current. The .second drogue 

 track (fig. 28) indicates that when the drogue 

 was launched the inertial current was flowing 

 with a northwestward component. The west- 

 ward movement of the drogue about one inertial 

 period later (1600Z on 13 August) supports this 

 drogue from the direct track between 2146Z on 

 12 August and 1600Z on 13 August represents 

 the diameter of the inertial circle, one can cal- 

 culate that the inertial velocity was 46 cm sec' 

 (Neumann and Pierson, 1966; p. 1.58). A similar 

 calculation on the drift from 1600Z on 13 August 

 to lllSZ on 14 August when the drogue was 

 recovered indicates that the inertial current liad 



decreased to 38 cm see"'. These speeds agree 

 with inertial speeds given by Pollard and Millard 

 (1970). 



It is of interest to speculate on the effect of 

 eddies such as this in exchanging water between 

 the slope and shelf areas. The average T-S 

 characteristics above 30 meters of stations on the 

 southwest side of the eddy are warmer and more 

 saline than those on the northeast (fig. 29). This 

 leads to the hypothesis that the anticyclonic 

 eddies are a contributing factor in the mixing 

 of shelf and slope waters. Another illustration 

 of possible eddy-related mixing in process can 

 be seen in the salinity profile of section A (fig. 

 13). The tongue of high salinity water found 

 between 10 and 30 meters on stations 4 and 5 

 suggests that an eddy can cause intrusion of 

 slope water onto the shelf. 



