During the second survey (fig. 20) it was again observed that 

 the minimum temperature of the Labrador Current was warmer 

 than usual and that the salinity of the intermediate water down 

 to 1,000 meters was fresher than usual. The use of the Helland- 

 Hansen and Nansen method of arriving at the dynamic heights of 

 stations where the depth of water is less than that of the reference 

 surface involves some uncertaintiy depending upon the construc- 

 tion of a vertical section of anomaly of specific volume for the 

 extrapolated values of the anomaly along the bottom from the 

 shallow station to the depth of the reference surface. The shallow 

 ends of such sections are cross checked by providing two or more 

 approaches to guard against gross errors in the construction of 

 the anomaly sections and to produce a consistent network survey. 

 In the construction of figure 20 such cross checks showed an ir- 

 reconcilable discrepancy of about 32 dynamic millimeters between 

 the two southernmost sections and the area to the northwestward. 

 The figure has been drawn with the discontinuity indicated by a 

 dashed line and it has been assumed that an adjustment between 

 two steady states occurred in the approximately 3-day interval 

 between the occupation of the southern sections and the stations 

 to the northwest of them. The current pattern is similar on each 

 side of the discontinuity but the shallow water velocities are 

 different. 



It would appear from figure 20 that an exceptionally large 

 part of the surface circulation of the Labrador Current in the 

 Bonavista triangle was following the eastern branch. An anal- 

 ysis of the volume transports past the three sides of the triangle, 

 however, indicates that 90 percent of the current was following 

 the eastern branch. This is only slightly more than the seasonal 

 normal of 88.3 percent. The seasonal normals, however, are based 

 on data which include only one occupation at a seasonal date 

 earlier than that of the second survey of 1958, and are conse- 

 quently less reliable than they are for somewhat later seasonal 

 dates. In the eastern part of the surveyed area figure 20 indicates 

 that the eastern branch of the Labrador Current suffered but 

 little loss from water recurving northward north of Flemish Cap. 



The dynamic topography found during the third survey is shown 

 in figure 21. As in other current charts of the Grand Banks region, 

 a reference surface of 1,000 decibars has been used in this figure 

 to permit the presentation of the extra-southerly section and the 

 rest of the survey in a single chart. It is realized that this ref- 

 erence surface is too shallow to give accurate values of current 

 velocity in the vicinity of the southern section but an inspection 

 of the data indicates the current pattern is similar when the 



39 



