310 PAPERS FROM THE DEPARTMENT OF MARINE BIOLOGY. 



If iV = chloride content of standard sea-water in grams per kilogram, 

 ^ = reading of burette with standard sea-water, and « = reading of burette 

 with sample, then N — A = a, which may be defined as the deviation of the 

 standard silver nitrate from the standard sea-water. Corrections k corre- 

 sponding to the values a at various concentrations a are then ascertained 

 from Knudsen's tables and are added to the reading of the burette in order 

 to obtain the correct content of chloride in grams per kilogram. 



The salinity S in grams per kilogram has been calculated by Knudsen 

 by means of the formula S = 0.030 + 1.8050 CI, and Knudsen's values 

 have been given for the calculated salinity corresponding to the content of 

 chloride of each sample. 



Similarly Knudsen's values for density at 0° C. referred to distilled 

 water at 4° C. (sq) have been given. His formula for calculating density is: 



So=i+-^^ 

 1000 



in which va is computed by the formula 



(To = — 0.069 -I- 1 .4708 CI — o.ooi57oCl^ +0.0000398C? 



ACCURACY OF RESULTS. 

 The discussion of the determinations has been limited to the period 

 September 12, 1914, to October 17, 1915, inclusive, as the hiatus in the 

 summer of 1914 makes it difficult to correlate the earlier estimates. The 

 average content of chloride for the 388 determinations, September 12, 

 1914, to October 17, 191 5, is 19.87 grams per kilogram. The average 

 deviation of individual observations is =^0.15. The deviation in 246 

 determinations is greater than the average, and in 131 determinations less 

 than the average, while in 11 determinations the deviation is zero. In a 

 series of 100 duplicate titrations, many on different days and by two 

 observers, the average deviation of individual observations is ±0.01 and 

 exceeds 0.02 in only 5 titrations, while the deviation of 32 titrations was 

 zero. The probable error of individual titrations certainly is less than 

 0.03 and the reasonable maximum deviation is less than 0.05. It is there- 

 fore reasonable to conclude, in view of the relative accuracy of the deter- 

 minations of chloride and the remarkable excess of small plus deviations, 

 that the sea-water at Fowey Rocks has a certain normal salinity, below 

 which it falls frequently because of certain recurring influences that result 

 in local intermittent dilution. The highest recorded contents of chloride 

 are 21.37 grams per kilogram (June 14) and 21.61 grams per kilogram 

 (June 16) during a period of heavy precipitation. These two estimates 

 exceed the others so markedly without apparent reason that some suspicion 

 is aroused regarding the samples, which were examined in triplicate. Only 

 one other estimate exceeds the average more than 0.30 and only 7 others 

 exceed it by more than 0.20 gram. On the other hand, 57 estimates are 

 less than the average by 0.15 or more, 27 by 0.30 or more, and 14 by more 

 than 0.60 gram. 



