90 PROF. WILSON ON THE ELECTRICAL CONDUCTIVITY AND LUMINOSITY, ETC. 
luminosity by an amount depending on the ratio of the concentration of the sodium to 
the concentration of the HC1. It appears that the reaction Na 2 0 + 2HC1 = 2NaCl + H 2 0 
does not proceed appreciably at the temperature of the flame. The small effect which 
the HC1 has on the luminosity is probably due to the change in the temperature of 
the flame together, possibly, with other causes. 
The effect of the HC1 on the conductivity was also independent of the ratio of the 
concentrations of the sodium and HC1. It follows, as for the luminosity, that the 
effect must be due to the change in the nature of the flame produced by the HC1 and 
not to chemical action of the HC1 on the sodium salts. Possibly the HC1 condenses 
on the negative ions, so diminishing their velocity. 
The effect of the HC1 on the fall of potential at the negative electrode may be due 
to its causing the electrode to emit negative electrons. 
In the earlier experiments of Smithells, Dawson, and the writer, it was found 
that at large concentrations oxysalts gave greater currents than haloid salts. The 
present experiments show that the conductivity is the same for haloid and oxysalts 
at all concentrations. It follows that the oxysalts diminish the fall of potential at 
the negative electrode when present in large quantities, while the haloid salts do not 
do so to the same extent. The amount of salt which entered the flame in the present 
experiments was rather greater than in the earlier experiments, for solutions of equal 
concentration. 
The present experiments on the variation of the conductivity with the concentration 
can be explained in a satisfactory way on the ionic theory, and they confirm the view 
that all salts give negative ions, having equal velocities in the flame. 
The fact that K 2 C0 3 does not diminish the luminosity due to Na 2 C0 3 seems to show 
conclusively that the luminosity is not due to positive ions. The K 2 C0 3 increased the 
number of negative ions about nine times, and, therefore, must have diminished the 
number of positive sodium ions in the same ratio. That the negative ions from 
K 2 COg do combine with the positive ions from Na 2 C0 3 is shown by the fact that the 
conductivity of a mixture of sodium and potassium carbonates is less than the sum of 
their separate conductivities as it should be according to the ionic theory. 
In conclusion, I wish to say that my thanks are due to the trustees of the Pice 
Institute for the facilities for experimental work which they have provided. 
