116 CARRITT [CHAP. 5 



This is usually used to compute salinity values from measured values of 

 chlorinity. Equation (1) has also been used as the definition of salinity. 



It should be noted that salinity as defined is not identical with total dissolved 

 salts. It represents a compromise between the desire to have a measure of total 

 dissolved salts without performing analyses for each individual constituent and 

 the limitations inherent in a less tedious set of procedures that gives a measure 

 of a property nearly proportional to the sum of individual constituents. Lyman 

 and Fleming (1940) discuss the difference in detail. 



Equation (1) is not consistent with the notion of the constancy of relative 

 proportions. Equation (1) predicts a salinity of 0.03% when the chlorinity is 

 zero, whereas the constancy of relative proportions would have them both zero. 

 Attention has been drawn to this problem several times in the past. For ex- 

 ample, Lyman and Fleming (1940), Carritt and Carpenter (1959) and Lyman 

 (1959) discuss various aspects of the problem, and it now seems clear that 

 neither the concept of the constancy of relative proportion nor the salinity- 

 chlorinity relationship, as expressed by (1), can be stated without modification 

 or qualification. 



Equation (1) was derived from the results of the analysis of nine surface 

 samples of which only four have a chlorinity in the range of 16% to 22 % . The 

 remaining five, Baltic and North Sea samples, were in the range 1% to 13% . 

 Lyman and Fleming (loc. cit., p. 143) noted that: "Brackish waters will vary 

 according to the nature of the particular river water with which they are 

 diluted. It happens that water from the Baltic region was selected as the 

 standard type of diluted water in oceanography, and comprehensive tables of 

 salinity, chlorinity and density relationships worked out. It follows from this 

 that it is not possible, by extrapolation to zero chlorinity of measurements 

 made in natural sea-water, to obtain the corresponding property of distilled 

 water, although this has been attempted in the case of refractive index, dis- 

 sociation constant, and other measurements. It is also incorrect to make 

 electrical measurements in sea-water over a range of chlorinity by diluting 

 a single sample with 'a good grade of conductivity water' although this 

 likewise has been done. Samples diluted with distilled water will not corre- 

 spond in properties to those of water of the same chlorinity from Knudsen's 

 tables". 



In terms of the physical, chemical and biological processes occurring in the 

 oceans and around their borders, constancy of relative proportions implies, as 

 noted by Maury (1855), that the oceans are "well shaken together". However, 

 if, as is the case of dilution with Baltic water, a decrease in chlorinity indicates 

 departure from "true oceanic properties" into regions where the ratios of 

 major constituents are not constant, how then do we define "true ocean water" ? 

 Dittmar (1884) quoting Forchhammer noted: "According to Forchhammer, 

 these ratios (ratios of the concentrations of the major dissolved constituents) 

 are, in passing from one part of the ocean to another, subject to only very 

 slight variations, if we omit the Caribbean, the German Ocean, the Baltic and 

 coast water generally". 



