442 
atmospheres. It will be shown presently that this value is too small. 
The oxide used in this tube was Sample A, which decomposes, as I have 
shown, far more slowly than B. It is probable that equilibrium had not 
been established when the experiment was ended. On account of this 
difficulty of determining when the reaction had come to equilibrium, the 
indirect method was abandoned and the following adopted: 
Two tubes, one having a bore of about 0.7 mm., the other of 3mm., 
were joined together and then drawn apart at the junction to a fine 
capillary about 14 m. long (BCD in Fig. 1). In the smaller tube D, 
a short column of mercury was introduced, and the open end sealed, thus 
making an ordinary closed manometer, which was calibrated in the usual 
way after the conclusion of the experiment. The tube B was filled with 
silver oxide and the open end sealed. 
Fia. 1. 
The large bath used in the preceding experiment had also to be aban- 
doned on account of the great consumption of dyphenylamine, of which 
only a limited supply was available. A simple bath consisting in a long- 
necked glass flask was found to answer the purpose. The neck of the 
flask was about 2 cm. in diameter and was insulated externally by a coat 
of asbestos. The temperature in different parts of the neck was frequently 
tested and the variation over a considerable range rarely amounted to half 
a degree. The tube B was held in a simple framework of glass which 
prevented it from touching the sides of the flask. 
In the preceding paper I have shown that the decomposition of silver 
oxide is extremely slow until a large amount of silver has formed, which 
acts as a catalyzing agent. Therefore, in order to hasten the attainment 
of equilibrium, the tube at the beginning of the experiment was heated to 
a higher temperature than that of the bath until a small part of the 
