LKWIS. — AUTOCATALYTIC DECOMPOSITION OF SILVER OXIDE. 729 



that the difference between samples was either due to varying quantities 

 of impurities or to variations in the physical state ; for example, in the 

 fineness of division of the powder. Pure silver oxide was therefore 

 prepared for study in three different ways. For all these preparations 

 silver nitrate purified by several re-crystallizations was used. In the first 

 case sodium carbonate was added to it. The j^recipitated carbonate was 

 washed repeatedly by decantation, the last washings occupying twenty- 

 four hours each. It was then dried and heated at 240° for twenty-four 

 hours, to drive off as completely as possible the water and the carbon 

 dioxide. The second and third preparations were made by treating the 

 silver nitrate with a clear solution of barium hydroxide, in one case with 

 both solutions at room temperature, in the other with both at 100°. All 

 the solutions were very dilute, and the reaction was carried on with the 

 exclusion of carbon dioxide. The ()recipitates were carefully washed, 

 and dried at 240°. 



It was expected that the silver carbonate would yield a very finely 

 divided oxide, and therefore one that would rapidly decompose, but as a 

 matter of fact it proved to react much more slowly than any previously 

 tried. The samples prepared with. barium hydroxide decomposed more 

 slowly still, — more than four times as slowly as the oxide used in the 

 earlier experiments. Contrary to expectation the oxide prepared in the 

 cold decomposed rather more slowly than the other. 



It would appear, therefore, from these few experiments, that the rapid 

 reactions of some of the samples was caused by the presence of im- 

 purities. Nevertheless my experiments were hardly sufficient to make 

 this conclusion certain, especially as certain experiments led me to suspect 

 that the rapidity and temperature of drying had some influence on the 

 rate of decomposition. Unfortunately, however, my experiments at this 

 point had to be brought to an abrupt close, and it is therefore necessary 

 to leave this question open. 



Temperature Coefficient of the Reaction Velocity. 

 The influence of temperature on the reaction velocity may be found 

 from tlie four experiments, the results of which are plotted in Figures 2 

 and 3, since these experiments were all made with the same sample of 

 oxide. We must choose for comparison the reaction velocities at corre- 

 Eponding points in the several experiments. The apices of the curves 

 constitute such a series of corresponding points. The maximum veloc- 

 ities at the four temperatures, 327.0°, 332.5°, 352.2°, 353.3°, were 

 respectively 0.0106, 0.0134, 0.0316, and 0.0335 per cent per second. 



