great constancy might be attained, for a thermostat working at 

 this temperature has two great advantages over the common water 

 thermostat — it is influenced very slightly l)y changes in the room 

 temperature and there is no loss of heat by evaporation. It must, 

 however, like all thermostats, be protected from drafts of air. 



The mixture of potassium and sodium nitrates proved to l)e 

 an ideal l)ath liquid. When pure it is as clear as water and as 

 mobile. Well-annealed glass apparatus and thermometers may be 

 plunged directly into the liquid again and again without l)reaking. 



The vessel of enameled ware seemed to suffer no deterioration 

 whatever, even after several months of contact with the molten salt. 



The apparatus for containing the silver oxide and measuring its 

 rate of decomposition is also sketched in fig. 1. The glass tube 

 (D), about 1 meter long, was filled with carefully dried silver 

 oxide to a height of' about 10 centimeters, and the powder was 

 shaken compactly into the end. This was covered with a thin 

 layer of glass wool and the remainder of the tube was Idled 

 witli a loose-fitting glass rod (F), the object of which was merely 

 to diminish as far as possible all air space subjected to changes of 

 temperature, and thus to avoid fluctuations in volume. The end 

 of the tube containing the oxide was immersed deeply in the ther- 

 mostat, and the other end Avas connected by rubber tul)ing to a 

 long brass tube (G), in passing through which the evolved oxygen 

 was cooled to the temperature of the room. The end of this brass 

 tube was connected with a gas burette (K) and to the recorder (L). 

 This latter was simply a glass tube about 2 millimeters in diam- 

 eter marked off at such intervals that the volume from one mark 

 to the next was exactly 1 cubic centimeter. A small colunm of 

 water (M) served as an index. With stopcock H closed and I 

 open the small water column advanced along the recording tube, 

 and the time taken by the forward meniscus in moving from one 

 mark to the next wa-< read with a stop watch. \Mien tlie reaction 

 was very fast, the recording meniscus was allowed to run tlirougli 

 several centimeters in order to permit a more accurati' measure- 

 ment of the time. Thus it was possible to measure accurately the 

 rate of evolution of oxygen, whether 1 cubic centimeter in an hour 

 or 1 cubic centimeter in ten seconds, the latter being the greatest 

 velocity measured. After determining the reaction velocity the 

 water indicator could be brought lia<U to its initial position l)y 

 cautiously opening the stoj)cock II. 1 was tin'ii closed an<l II 



