70 A Retrospect of Oceanography 



be comparable when they have not been made at the same 

 temperature, it is necessary to effect some reduction. It may 

 be taken that the variation of density of distilled water with 

 change of temperature has been determined with the greatest 

 attainable accuracy. If we multiply the specific gravity found 

 by the density of distilled water at the temperature, the result 

 will be the density of the sample at the same temperature; 

 or, as it may be otherwise expressed, the specific gravity 

 referred to that of distilled water at 4 C. as unity. This 

 value is affected only by whatever error may be inherent 

 in the number taken for the density of distilled water. 



To make further reductions, we require a knowledge of the 

 law of thermal expansion of the sea-water. This cannot be 

 known with the same precision as that of distilled water, 

 consequently its use should be kept within the narrowest 

 possible limits. The density of a sea-water is determined 

 for two perfectly distinct purposes the one is chemical, in 

 order to obtain a knowledge of its salinity; and the other 

 is physical or mechanical, in order to know its effect on the 

 equilibrium of the waters. In order to compare waters with 

 regard to their salinity, their densities must be reduced to their 

 value at a common temperature. Such reduced densities 

 give a very accurate representation of the relative salinity 

 of different waters. In order to obtain absolute values, resort 

 must be had to one of the tables connecting these two variables. 

 For mechanical purposes, it is requisite to know the density 

 at the temperature which the water had in situ. A further 

 reduction might be made to allow for the effect of pressure 

 in the case of water collected below the surface, but it is rarely 

 of any mechanical importance. 



It is well to arrange beforehand to determine the specific 

 gravities of the waters during a cruise at as nearly as possible 

 the same temperature ; the results are then comparable, and with 

 the minimum of reduction. It is my practice, in such circum- 

 stances, to take the mean of the temperatures at which observa- 

 tions have been made, and reduce all the observations to their 

 value at this temperature. This gives the best account of 

 their relative salinities amongst each other. To make them 



