306 MR. W. R. BOUSFIELD AND DR. T. M. LOWRY ON THE ELECTRICAL 



We were thus enabled, for the first time as we believe, to express the relation 

 between the temperature and conductivity of an electrolyte in terms which exhibit 

 the separate effect of the two main factors involved, namely, change of viscosity and 

 change of ionisation. 



It now remained to determine the variation of the coefficients over the whole range 

 of concentrations. In the present communication we have been obliged to limit our 

 observations to the range of concentrations from 50 per cent, down to 4 per cent, 

 (normal), but we hope at some future time to resume the investigation of the more dilute 

 solutions, which promise to yield results at least equal in importance to those derived 

 from the study of the more concentrated solutions. The values of the coefficients are 

 somewhat irregular if derived either by the method described above or by calculation 

 from four particular observations of any series, since small experimental errors might 

 produce much larger errors in the values of the coefficients. The points chosen for 

 the initial computation of the coefficients were the experimental values for K , K^ 

 and K 100 for each of the concentrations studied. On plotting out the values so 

 obtained it appeared that the values for a and n could be very simply expressed in 

 terms of the concentration. The relation of n to the concentration m of the solution 

 was n = I'910+log 10 (m+0'654) 2 , and we were able to adjust the constants in this 

 expression so that when m = 0, n = 1'542, the index given by THORPE and RODGER 

 for the index n in SLOTTE'S formula for the fluidity of water. This relation, we have 

 reason to believe, holds good right through the range of dilute solutions down to 

 water. 



From 50 per cent, nearly down to normal the coefficient a is related in a linear manner 

 to the concentration m, a result that may be expressed by means of the equation 

 a = 0'00293-f-j^oo m - 1 the case of "water" we have reason to believe that there 

 is no appreciable "decay of ionisation" between 0C. and 100 C., so that when m=0 

 we should have a = 0. The rapid drop in the value of , which actually occurs 

 below normal concentration, is not expressed by the above linear relationship, and 

 we have provisionally added a small term to the linear equation, which is then 

 made to read a = 0-00293+: ro 1 -oW 0'000146/(m+-^ -). The effect of this added 

 term is to make a = when m = 0. We believe that further investigation will 

 justify some such addition, which hardly affects the figures for concentrations above 

 normal, and which corresponds roughly to the facts indicated by our incomplete 

 experiments on dilute solutions. 



Having thus determined the relations of n and a to the concentration from a study 

 of all the individual values for K , K 50 , K 100 , we now took the smooth values for n and a 

 given by the above expressions, and recalculated the values of & from the equation 

 logK = logK -f ttlog(l + 6<) at, by substituting the experimental values of K 100 

 and K , and the smooth values for n and a. The same process was repeated using 

 the experimental values of K 50 in place of those of K 100 . In this way all the errors 

 of observation, whether from inaccurate reading or fouling of solutions or determina- 



