SECT. 1] CHEMICAL INSTKUMENTATION 117 



It is, then, perhaps surprising to note that, despite the warning noted above, 

 approximately half of the samples that were analyzed for chlorinity and 

 density and the results used in the construction of Knudsen Tables, and 

 approximately half of the samples that form the basis for equation (1) 

 were taken where departures from constancy of relative proportions would be 

 expected. 



With only nine measurements of salinity, made on a set of samples which 

 most certainly are not representative of all oceans and all depths, it is natural 

 to question the applicability of the relationships based upon those samples to 

 all ocean waters. How different will the actual salinity be from that computed 

 from (1) and a measured value of chlorinity? On the average the difference 

 will probably be small. Lyman and Fleming (1940) computed the sum of major 

 anions and cations from the average of measured element-to-chloride ratios and, 

 after applying the necessary corrections to obtain salinity as defined, obtained 

 a value in excellent agreement with that computed from (1). This, however, is 

 only a part of the problem, because we are seldom concerned with average 

 conditions. Often we depend upon departures from a mean to provide informa- 

 tion about processes occurring in the ocean. 



Obviously it is possible to resolve these difficulties. We need only collect a 

 representative set of samples and measure salinity (as defined) and chlorinity. 

 At this point, however, it seems hardly worth the time, effort and expense. 

 Salinity is a denned property of sea-water and one that is difficult to measure 

 with high accuracy and precision. Furthermore, the use of computed values of 

 salinity, even though we had reliable empirical relationships, tells us nothing more 

 about the oceans than would be obtained by using directly the property from which 

 we compute salinity. For example, we can learn no more from a temperature- 

 salinity curve or envelope than from a temperature-chlorinity envelope, 

 and the latter is fundamentally a truer statement of the information at 

 hand. 



If it is desirable to retain the concept of salinity, that is a property which is 

 roughly proportional to the total dissolved salts, other properties that can be 

 readily measured with high accuracy and precision can be used. For example, 

 the total anion or cation equivalents can be measured by ion-exchange tech- 

 niques which terminate in a simple acid or base titration. These measures 

 will be more nearly proportional to total dissolved salts than any of the proper- 

 ties now measured and linked empirically with chlorinity, and so to salinity. 

 For example, conductivity and refractive index give measures of the sum of 

 the property of each of the dissolved constituents in sea-water. However, 

 neither of these properties is strictly analogous to salinity. The measured 

 conductance of a sea-water sample can be written as the sum of the ion con- 

 ductances of each of the dissolved constituents. However, the ion conductance 

 of one constituent is not a simple function of the concentration of that con- 

 stituent. Thus, it is possible to have significant changes in the concentrations 

 of, say, two constituents, as in an exchange process, and so significant changes 

 in salinity and yet insignificant changes in conductance. 



