AND THE CUBIC COMPRESSIBILITY OF MERCURY. 47 



wrapped with a piece of thin silk, except above. In this manner it was 

 possible to insulate the rod perfectly, if the fluid transmitting the pressure 

 was a non-conductor. 



Limpid paraffin oil was used to convey the pressure, which was applied 

 through the nut H at the bottom. Amagat used water, but in the present 

 experiments it was found impossible to insulate adequately the metal rod 

 if the steel cylinder contained this liquid. 



The method of procedure was as follows: After mounting, the steel 

 cylinder E and the metal rod R were respectively connected with the two 

 poles of a weak galvanic cell, one side of the circuit containing a delicate 

 index-galvanometer. All being finally adjusted, the bolt C was turned 

 until the platinum point just touched the surface of the mercury in the 

 cup at the top of the metal rod. The platinum point was carefully cen- 

 tered, so that in the act of turning it remained exactly in the middle of the 

 tube, no matter whether bolt C was screwed in or out. The moment of 

 contact between the point and the mercury was indicated by the gal- 

 vanometer. 



The index A was tightly fitted on the rod K, exactly above the top of 

 the needle B. Pressure was now applied, until the circuit was broken 

 because the metal rod had been shortened so much that the platinum point 

 no longer touched the mercury. The pressure was allowed to fall slowly 

 until the galvanometer showed once more a completed circuit. The press- 

 ure at the moment of making the circuit was recorded, and this process 

 was repeated at least twenty times. It was the custom to record ten of 

 these pressures on contact, then to release the pressure, either completely 

 or partially, and to apply it again after a short interval, and repeat the 

 whole process. The average of these twenty readings was taken as the 

 pressure required to make the contact. The distance between the index 

 A and B was now read by means of a very accurate telescope micrometer. 

 During the whole time the water in the jacket was stirred. This served 

 to distribute the heat of compression and to eliminate it as soon as 

 possible. 



The bolt C with its needle B was now turned so as to lower the point 

 about 0.15 mm., a change which caused the contact to hold until about in 

 the neighborhood of 500 atmospheres. The same process was repeated, 

 the pressure of contact being recorded and the distance between A and B 

 read as before. At the high pressures the pressure was not suddenly 

 released between the readings, since this would cause the rod to shorten 

 from cooling. Instead, the pressure, after being allowed to fall slowly 

 over a range of about 200 atmospheres, was increased slowly before the 

 second series of readings was taken. For the most part the two series 

 agreed satisfactorily. From the data it is clearly possible to calculate how 



