108 MESSRS. A. SMITIIEU.S, II. M. PAWsOX, AND H. A. WILSON* : F.I.KCTIIICAI. 



sodium salts, experiments at great dilutions were not attempted with them. The 

 1,'ivatest dilution at which all salts were examined was -fa normal, and, although at 

 this concentration the various salts of each metal do not conduct equally well. \\c 

 may take the chloride as representing the haloid salts and the nitrate as representing 

 the oxysalts. The following tahle gives the comparison : 



It is evident that the conductivity increases with increasing atomic weight of the 

 metal, and that the increase is more rapid in the case of the oxysalts than in that of 

 the haloids. 



Influence of Temperature on Conductivity. 



In the form of apparatus used by us a column of hot gases ascends from the lower 

 cone of combustion to the upper one, the temperature rapidly decreasing. By altering 

 the relative position of the electrodes and the mouth of the flame tube the electrodes 

 could be brought into regions of different temperature. We do not attempt in the 

 present enquiry to deal fully with the relation between the temperature of salt 

 vapours and their conductivity, but have contented ourselves with making a few 

 experiments, so as to gain an idea of the general order of the relationship. 



By means of two millimetre scales etched on opposite sides of the glass cylinder, G 

 (fig. 3), the distance between the tip of the inner cone and the lower edge of the 

 electrodes (which in these experiments were square plates of platinum) was adjusted 

 to 5, 15, and 25 millims. in the respective cases. 



To measure the temperature, a platinum platinum-rhodium thermo-couple was 

 placed first 2 millims. above, then 2 millims. below the electrodes in the axis of the 

 flame, and the mean of the galvanometer readings taken to represent the temperature 

 of the vapour between the electrodes. The solution sprayed in all cases was normal 

 sodium carbonate. 



The following table contains the results; the observations were repeated in the 

 order given, so as to control their accuracy : 



