DETERMINATION OF SALINITY 



71 



the date of measurement to determine whether or not 

 there had been any change in the bridge calibration 

 which might have been caused by differential aging 

 among the end coils and slide wire or by corrosion. As 

 the differences were not systematic with respect to 

 time, no correct'ons for time were applied. 



These differences were then plotted against their 

 respective salinities as given by equation (2). In this 

 case systematic differences were found and a smooth 

 curve drawn through them. The departures of this curve 

 from zero were then tabulated as corrections to be ap- 

 plied to the salinities determined from the slide-wire 

 readings by means of equation (2). These corrections, 

 given in table 2, are largely attributable to irregulari- 

 ties in the slide wire. The alternative assumption is 

 that these differences arise from variation in composi- 

 tion of the salt, and such an assumption would require 

 that the composition be an irregular function of the sa- 

 linity. Such a relation seems highly improbable. When 

 these corrections are applied to the salinities derived 

 from equation (2), of the 219 comparisons, 212 differ 

 from the titration values by amounts equal to or less 

 than 0.04 per mille salinity. This seens to show an even 

 greater constancy of salt composition than has been as- 

 sumed in the past and leads one to question the accuracy 

 of chemical analyses of sea water as published in the 

 past. Such published analyses indicate that if solutions 

 of each were adjusted to equal concentration, the salin- 

 ities as given by titration would differ in some cases in 



Table 2. Corrections to be applied to 



salinities computed from 



second degree equation 



the first decimal place of parts per thousand. Obviously 

 no such variations were encountered in the cruise of the 

 Carnegie . 



Asa routine matter the samples of sea water col- 

 lected at an oceanographic station in the morning were 

 transferred, on arrival at the surface, to glass bottles 

 of the citrate-of-magnesia type. These bottles were of 

 about 350-cc capacity and were equipped with patent rub- 

 ber washer stoppers. The same glass bottles were used 

 repeatedly and were used only for sea water. The rub- 

 ber washers were replaced as often as their deteriora- 

 tion required. To guard further against evaporation, di- 

 lution, or contamination of the samples, their salinities 

 were measured on the afternoon of the same day they 

 were collected. 



The covers were removed from the salinity bridge 

 and the stirring motor and thermostatically controlled 

 heaters of the water bath were set in operation about an 

 hour before measurements were started, in order to 

 have equilibrium conditions of temperature established. 

 The salinity bridge had three measuring cells, any one 

 of which could be switched into the bridge circuit. The 

 auxiliary cell had in series with it a small adjustable 

 wire-wound resistance which will be called Q for con- 

 venience. The first step was to exhaust the measuring 

 cells of the water which had been standing in them and 

 to fill them with standard water after rinsing them with 

 some of the same standard water. The sealed glass tubes 

 of Copenhagen standard water were opened only as need- 

 ed and their contents transferred to a glass-stoppered 

 stock bottle. Any standard water remaining in the stock 

 bottle from a previous run was used only for rinsing, 

 the cells being filled with water from newly opened tubes. 

 The time of filling each cell was then recorded. Fifteen 

 minutes after the first cell was filled with standard 

 water, the slide wire was set at a reading corresponding 

 to the salinity of the standard water and the bridge bal- 

 anced by adjusting Q. This adjustment was then tested 

 by moving the slide wire and rebalancing the bridge by 

 adjusting the slide wire. The setting of Q was correct 

 if the slide wire was brought back to its original setting 

 to balance the bridge. This reading of Q was then re- 

 corded for this particular cell, the standard water was 

 removed from the cell which was then rinsed and filled 

 with water from one of the samples to be measured. 

 Re",ord was made of the time of filling and the identity 

 of the sample which took its designation from the num- 

 ber on the glass bottle in which it had been stored. Then 

 the adjustment of Q was determined for the second cell, 

 which in turn was filled with water to be tested. A sim- 

 ilar procedure was followed for the third cell. Fifteen 

 minutes after the unknown was placed in the first cell, 

 Q was set to the reading previously obtained for thrt 

 cell, the cell was switched into the circuit, and the bridge 

 balanced by the adjustment of the slide wire. This slide- 

 wire reading was recorded and the sample withdrawn 

 from the cell, which was then rinsed and filled with water 

 from another sample. The time of fillijig was again re- 

 corded and similar operations performed with the other 

 cells until all the samples had been measured. As the 

 last sample in each cell was removed, it was replaced 

 by standard water. This standard water was measured 

 exactly as if it were an unknown sample. The difference 

 between the slide-wire reading for this final standard 

 and its correct value (that for the initial standard) rep- 

 resented changes in the cells, changes in the solution in 

 the auxiliary cell, or errors in the measurement of 



