236 



Journal of the Asiatic Society of Bengal. [June, 1914. 



These observations made it clear that moisture gradually 

 entered the case, so that the sulphuric acid became more and 

 more dilute until it ceased to produce any effect. It was 

 thought that this moisture probably entered by diffusion through 

 the teak, so that this was soaked in paraffin to protect it from 

 direct contact with moist air. This merely retarded the deterio- 

 ration of the insulation. A leak of 0-01 volt per minute from 

 the quadrants was not reached until sixty hours after the case 

 was closed. These measurements, made from September 11th 



to 16th, are shown in the diagrams* 



In order to test whether diffusion of water vapour through 

 the wood was the cause of the failure of the insulation, the 

 whole of the case was protected with a zinc cover, soldered at 

 all edges so that moisture could only reach the interior through 

 the small holes left for the passage of the threads to the key. Ob- 

 servations made after this alteration showed that a satisfactory 

 solution had been obtained, as although the wood was damp 

 when first enclosed the leak of the quadrants remained below 

 0-01 volt per minute for nine days after the case was soldered 

 up. For the first sixty hours the measurements are shown in 

 Table V : 



Table V. 



Drying agent, sulphuric acid. Sealed zinc case. September 



18th to 26th. 



Leak after in 



Hours after 

 closing. 



NG (VOLTS PER MINUTE). 



Quadrants only. 





4 



12 



Quadrants and 

 sulphur plug. 



Quadrants and 

 condenser. 



24 



3G 



48 



0-0018 



0-0027 



0-0044 

 0-003 



0-0048 



0-0081 



0058 





0052 



60 



0-0055 



0034 



0055 



0047 



0020 

 0-0012 

 0O014 

 0-0014 

 0-0015 

 0-0018 



When the wood inside the zinc cover had become thoroughly dry, 

 the improvement of the insulation became still greater. In the 

 final series of observations, the first part of which is shown 

 in Table VI, the quadrant leak remained below the standard 

 value for more than fifteen days, although the weather was very 

 unfavourable owing to heavv rain from October 13th to 16th. 



