522 
MR. H. A. WILSOX OX THE ELECTRICAL COXDUCTIYITY 
tions of the velocity of the blast greater than 20 per cent., the necessary P.D. 
diminished more rapidly than the velocity of the blast. This vas j^robahly due to 
the effects of accidental air currents and diffusion, Avhich, of course, become more 
important when the blast of air is made less rapid. If the velocity of the blast was 
reduced to mucli less than one-half its usual value, the grating did not become hot at 
all, and practically no current at all could be obtained. 
Any change in the temperature of the platinum tube produced a change in the 
amount of current obtained with any given P.D., and also changed the value of the 
least P.D. necessary to give an appreciable current. 
The gas and air supplies to the gas furnace were therefore carefully regulated by 
means of the apparatus already described for producing a steady flame. In this way 
the temperature of the tube could be maintained sufficiently constant for any length 
of time. 
Increasing the temperature increased the amount of current obtained, and diminished 
slightly the least necessary P.D. 
Since, however, altering the temperature of the tube alters so many of the other 
conditions of the experiments, such as the velocity of the air blast and the rate of 
volatilisation of the salt bead, it is very difficult to interpret the meaning of this last 
result. Consequently in all the experiments the temperature of the tube was kept as 
nearly as possible the same. 
The accuracy with which tlie necessary P.D. could be determined varied with the 
distance between the electrodes and with the temj^erature of the tube. If the 
electrodes were too near together, a current was obtained with any P.D., however 
small. 
Increasing the temperature diminished the least distance at which satisfactory 
results could be obtained. 
Finally, a distance of 1 centim. between the electrodes and a definite supply of 
gas and air to the furnace and to the tube were adopted, as giving the best results. 
The P.D. necessary to produce an appreciable current could be determined within 
1 volt in any one experiment, and the results of different experiments with the same 
salt usually agreed within the same limit. 
Occasionally larger discrepancies than this occurred, which, however, could generally 
be traced to some accidental circumstance affecting the supply of gas or air to the 
furnace. 
The amount of current obtained with any one salt depended much less than might 
be expected on the size of the bead used. With some very volatile salts such as KI, 
a bead about 3 millims. in diameter only lasted two or three minutes, yet during 
nearly all this time the current with a given P.D. remained very nearly constant 
and only began to fall off when the bead had almost disappeared. This, of course, 
greatly facilitated the measurements, since it was not neaessary to keep the bead very 
constant in size durini>’ a series of determinations of the current with different P.D.’s. 
