ON THE SPECIFIC HEAT OF WATER. 235 



From these figures we obtain for the mean value of J from C. to any tempera- 

 ture 6 the values set out in Table IX. 



TABLE IX. Values of J from C. to 6. 



e. J'. 



C. 



13 4-1937 



27 4-1841 



40 4-1813 



55 4-1848 



73 4*1892 



80 4-1913 



An important test of our methods of working and of computing the results is 

 obtained by working out the figures for the mean value of J for the interval from 

 13 C. to 55 C. from the figures for the intervals from 13 C. to 27 C., 27 C. 

 to 40 C., and 40 C. to 55 C. now obtained. By this process we arrive at the result 



J,, M = 4-1819. 



The figure, at which we arrived by the series of direct experiments over the 

 complete interval, without any breaks, was 



J 13 M = 4-1821. 



The two sets of experiments were entirely independent, and the close agreement 

 indicates that the various small errors of observation which occur in obtaining the 

 time and temperature values at the beginning and end of short intervals, and the 

 small errors of computation, which result from taking the mean values for current and 

 resistance as the values at the middle points of the intervals, all average out in such a 

 way as not seriously to impair the accuracy of the results. 



14. Deduction of Values of J for Various Temperatures. The values of J probably 

 require a curve of the fourth degree for their accurate expression, or even of the fifth 

 degree over the range of C. to 100 C. This would be in accord with the expression 

 required for the density of water from C. to 100 C., which can be accurately 

 represented by an expression of the fifth degree. It is probable that liquid water is 

 a mixture of three species of molecules, to which SUTHERLAND has given the useful 

 names hydrol (H a (), or steam molecules), dihydrol (H 4 O), and trihydrol (H 8 O S , or ice- 

 molecules) (see " Liquid Water a Ternary Mixture," BOUSFIELD and LOWRY, 'Trans. 

 Faraday Soc.,' vol. G). It is further probable that both the specific heat aud the 

 drnsity of liquid water would depend additively upon the specific heat and density of 

 their components, and hence that if an equation of the fifth degree represented the 

 density of water, an equation of the same degree would be required for the specific 



2 H 2 



