730 Mr. A. E. Garrett on Electrical 



50-60 divisions for one volt. Hence it is quite possible that 

 some of those cases in which no increased conductivity could 

 be detected by such means, did produce an increase in the 

 conductivity, but it was too small to be detected. JE. g„ in 

 some later experiments, when using a Dolezalek electrometer 

 giving a deflexion of 600 scale-divisions for one volt, a con- 

 siderable leak was obtained by heating lead iodide. It will 

 be seen from the preceding list that the halogen salts of some 

 metals when heated make the surrounding gas a conductor 

 of electricity. 



As no attempt at this stage was made to obtain anhydrous 

 salts, the presence of water may have altered their ability to 

 undergo chemical change, and in so doing may have altered 

 their powers of producing conductivity. 



Many of the chlorides which were found to be able to 

 produce a leak of electricity are known to undergo chemical 

 change when heated. It is probable that, at the moment 

 chemical change takes place, an emission of an electrified 

 particle occurs. In the case of the nitrates, the clean iron of 

 the tube in which these compounds were heated was oxidized, 

 and Beattie * and Schuster have both shown that this process 

 causes the surrounding air to be positively electrified. As, 

 however, in the case of ammonium nitrate an increased 

 conductivity was observed when the electrode C was 

 positively charged, this does not account for all the results 

 obtained with those compounds. 



As a general rule, the leak was larger when due to positive 

 ions, and that leak could be detected in most cases at lower 

 temperatures than when due to negative ions. 



Later, some zinc phosphate was tested by heating up to 

 290° C. at a pressure of 10 mm. of mercury. Although the 

 electrometer gave a deflexion of 700 divisions for one volt, no 

 leak was detected. Potassium iodide was also tested up to 

 temperature 260° C. at a pressure of 10 mm. of mercury. 

 The electrometer gave a deflexion of 600 for one volt, but 

 no leak was detected. 



Thallium iodide was heated up to 210° C. at a pressure of 

 about 15 mm. of mercury. The electrometer gave a deflexion 

 of 350 for one volt, but no leak was detected. 



The negative result with potassium iodide is of interest, as 

 H. A. "Wilson in his papers f on conduction by salt vapours 

 when sprayed in solution into a hot tube, found conduction y 

 even when measuring currents with a galvanometer, at 

 temperatures near 270° C. (It is shown below that the leaks 



* South African Association, vol. i. April 1903, p. 4, 

 t Phil. Trails. A. p. 296 (1901). 



