difference between the corrected readings of a pair of thermometers of 

 ±.018° C. It is considered then that the observed temperatures, which 

 for the most part are the means of the corrected readings of a pair of 

 thermometers, are accurate to about ±.01°. 



Nansen-type water bottles were used for the collection of water samples. 

 As in previous years routine determinations of salinity were made by the 

 electrical conductivity method using a Wenner salinity bridge. As the 

 bridge had not been in use since the end of the 1941 season it was com- 

 pletely overhauled during February, 1948. The third or Wagner arm 

 shunting the ordinary Wheatstone circuit had been made up of two 

 fixed end-coils of about 500 ohms each between which was a 100-ohm 

 slide-wire shunted by a fixed impedance of about 20 ohms. The slider 

 of the slide-wire was connected to ground during its adjustment to balance 

 in bringing the electrical center of the bridge to ground potential. In 

 order to increase the range of the salinity samples which could be accom- 

 modated under this condition, the effective proportion of the third arm 

 which would be varied by the slide-wire was increased by shunting each 

 of the end coils with a 1000-ohm impedance and replacing the shunt 

 across the slide wire with a 100-ohm impedance. This changed the 

 overall impedance of the third arm from about 1019 ohms to about 718 

 ohms. The fixed impedance in the X-dial branch of the bridge was 

 measured to be 200.1 ohms. 



Fourteen samples of sea water, about evenly distributed over the range 

 from 30.44 to 36.69%o salinity were then compared in the bridge against 

 Copenhagen standard water of the batch Pi 5, using a bridge X-dial read- 

 ing of 50.040 as corresponding to the salinity of the standard water. 

 The salinities of the fourteen samples of sea water were determined by 

 silver-nitrate titration. A calibration curve of the form S = A/(B -\- m) 

 — C was then derived were S is the salinity in parts per thousand, m is 

 the reading of the X-dials with the bridge balanced, and A, B, and C 

 are constants of the bridge. B was taken as 200.1 from the measurement 

 made with an external bridge. All combinations of the 14 sets of data 

 gave 91 individual values for the constant C from the relationship, 



(.Somo - Smi) - 200.1 (>S'i - ^2) 

 (mi - 7)12) 



The weighted mean of these values was derived by dividing the sum of 

 the numerators by the sum of the denominators. This weighted mean 

 was 3.961 which was used to substitute in the expression, 



A = (Si + C) (200.1 + mi) 



giving 14 values of A whose average was 9754.066. The resulting ex- 

 pression for the cahbration curve, S = 9754.066/(200.1 + m) - 3.961, 

 was then used to check the departure of the titration values from the 

 curve. For the 14 points the departure averaged — 0.002 ± 0.023%o. 



72 



