in the Twenty Years before 1895 71 



comparable with other observations, they must be reduced 

 to one of the usual common temperatures, as, for instance, 

 I 5'56 C. (60 Fahr.). It will be remembered that we are 

 dealing here with densities, that is, the weight in grammes of 

 i cubic centimetre of the liquid, and if we reduce our densities 

 to 15-56 C., we shall have the weight of a cubic centimetre 

 of the water at that temperature. In some countries, especially 

 in Germany, it has been the custom to reduce results to their 

 values at 17-5 C. referred to the density of distilled water 

 at the same temperature. The temperature 17-5 C. is not 

 open to any objection, but we ought to adhere to densities, 

 where our unit is the weight of a cubic centimetre of distilled 

 water at 4 C. The common temperature to which the 

 densities are to be reduced is, in itself, not a matter of any 

 importance. The consideration which should guide us in 

 its choice is to reduce to a minimum the alteration produced 

 in the observed value by the reduction. As the usual tem- 

 perature at which observations are made is the atmospheric 

 temperature of the locality, we have to consider what is likely 

 to be the most frequently occurring temperature. Many more 

 observations are made in warm seas than in cold ones, and 

 the area of the warm seas is greater than that of the cold 

 ones. It would then fore seem reasonable to select a high 



perature rather than a low one. 



A sea-water of density 1-02600 at 15-56 has the following 

 at other temperatures: 



As observations must always be made in a place protected 



currents of air. it may be taken that 3 C. will be as low 



a temperature of observation U il likely to occur. Similarly, 



29-8 C. may be taken as a higher extreme. At 3 C. the above 



water has a density of 1-02800, and at 29'8 C. its density 



02200, tin dittrrence being 0-00600; the mean of these 



1-02500, occurs at 19*6 C., and if we assume that ob?< 



