358 Mr. G. F. Emery. [Apr. 19, 



to give accurate readings all through and enable me to get a curve 

 showing the connection between E.M.F. and difference of tempera- 

 ture. In every case this was very nearly a straight line, and most of 

 the experiments were made pretty quickly so as to get accurate 

 ings for the highest and lowest temperatures, while those for int 

 mediate points served to show if any irregularity had taken pi 

 from secondary causes, and form a check on the accuracy of the whc 

 If the readings are not steady or differ much for corresponding poii 

 on rising and falling temperatures, I do not, as a rule, attach any 

 value to that experiment. A change of ^ C. in the temperature of 

 one of the junctions gave an easily noticeable deflection to the 

 vanometer spot. 



For convenience I have used the symbol for the electromot 

 force per 1 C., and the unit in terms of which 3 is given is 10" 4 

 per 1 C. 



The value of d9[dt is in all cases very small compared with tl 

 3, and appears to vary in sign for different solutions. 



Assuming, then, that dSjdt = 0, as is approximately the 

 my object was to find the variation of with the strength of 

 solution. 



I made a long series of experiments with six salts, acetate, chloi 

 and sulphate of zinc, and sulphate, nitrate, and acetate of cop{ 

 Nitrate of zinc and chloride of copper gave no results as they appeal 

 to attack the electrodes. In a whole series of experiments on 

 salt I used the same lot of boiled distilled water and the same si 

 solution, which was diluted down to the required strength. I thir 

 therefore, that any small effect due to impurities in the water may 

 safely taken to be a constant. The salts were all as pure as I cot 

 obtain, and the stock solution was kept for a long time over oxide 

 carbonate so as to neutralise any free acid as far as was possible. 



I found considerable variations in the value of when the concent 

 tion of the solution was varied, and to show these I have plotted cm 

 (fig. 2), whose ordinates are equal to , and abscissae are the coi 

 spending concentrations in gram molecules per litre of volume, 

 very small concentrations it seems impossible to observe accurat 

 and the parts of the curves corresponding to very small concent 

 tions I have extra-polated as well as I can. For zero concent 

 the curves all appear to start from somewhere about # = 8'6 at 

 angle to the axes, and then as concentration increases the curves 

 round more or less sharply till they are nearly parallel to the axis 

 concentration. It will also be observed that for some salts incr 

 with increasing concentration, while for others it decreases, so 

 for thermo-electrical purposes we may divide salts into positive an' 

 negative according as the value of 3 for the solution is greater or les 

 than its apparent value for pure water. In all the aqueous solution 



