144 



Values for m and n found in testing tlie apparatus in February, 1923, are 

 as follows : 



Hi=(».00012r) M=0.000143 



From the data published by Thuras we find in the case 8' is taken as 

 30 parts per thousand. 



-l = — 0.02847 7;==0.00064S 



These values for .1, L'. m, and n. give — 



« =0.009413 

 ^=— 0.000001344 

 r=0.0000000001920 

 rf =0.002276 



Appendix II 



SUGGESTIONS FOR THE CoXSTKrCTIOX OF NEW CONVEKSION TABLE OR CURVE 



In order to avoid all possibility of errors hi the determinations made by 

 Thuras. in the calculation of the constants A and B from his data, errors in 

 the test of the apparatus or possible changes in the apparatus, and in the cal- 

 culation of the table it may be advisable to construct a new table or curve 

 based directly upon measurements made with the apparatus, using samples of 

 sea water who.se salinity has been determined by the titration method. The 

 table given above is for sea water of salinities from 30 to 38 parts per thou- 

 sand, and might have been extended to about 40 parts per thousand before 

 reaching the upper limit of the scale. In making a new table or curve, it 

 seems that it would be well to start with a reading of about — 

 200 for a salinity of 30. 

 500 for a salinity of 33. 

 600 for a salinity of 34. 



This would make the range of salinities whicli might be measured from 

 about 28 to about 38. In making a new table or curve, as in the use of the 

 apparatus and as in the determination of salinities by the titration method, 

 it is necessary to have a quantity of sea water of known salinity to be u.sed as 

 a standard. However, this standard need not be the same as that which will 

 later be taken aboard the ship. 



With these values of a. h. c. and (/ substituted in eciuation 7. the data is in 

 about as convenient form for calculations as it can conveniently be put. How- 

 ever, it would not be practicable to calculate the salinity directly from this 

 equation each time a measurement were n)ade. Therefore the equation was 

 used in the construction of the table given above which is for the purpose of 

 converting the readings of the apparatus to salinities. This table gives .salini- 

 ties directly for readings of 0. 10. 20, etc., up to a reading corresponding to 

 a salinity of 38 parts per thou.sand. It is so tabulated that interpolations can 

 readily be made to 1 in the reading of the slide wire or to 0.01 in the salinity. 



Let us suppose that the sea water to be taken as a standard for this purpose 

 hiis a salinty of 29.75. Then to have the table or curve extend over the range 

 suggested above we should set the slide wire index at a reading of 175 and this 

 arl)itrariiy fixes one point <>n the curve. Other points on the curve are then 

 fixed with respect to this one when the ciianges in reading are noted as the 

 sea water of salinity 29.75 in the T cell is replaced with sea water of other 

 Icnown salinities. 



The greater the number of points from whidi to plot the curs-e the more 

 r(>lla))!e it should be. However, thei'e is every reason to expect the curve to be 



