Measuring very High Pressures. 453 



but with the connections described above 

 R -R = p(x-x) 



hence 



a ±v„ 



and all the terms except P and x are constant so long as the 

 temperature remains the same. During the present investiga- 

 tion the cylinder containing the mercury tube was surrounded 

 by melting ice nearly an hour before the observations were 

 begun in order that the temperature might become uniform, 

 and after each increment of pressure sufficient time was 

 allowed for the heat of compression to escape. The variation 

 in the temperature of the wire was never more than two 

 degrees and the change in j3 was negligible. The value of a 

 at 0° C. is -00003237, and R was so chosen that / 3 /aR = 4*09. 

 Hence 



P = 4-09(x —x) 



and, since a motion of the contact on the wire equal to two- 

 tenths of a division was clearly indicated by the* galvanometer, 

 a change of pressure of one atmosphere could be easily 

 detected. 



The apparatus illustrated in fig. 1 was designed primarily to 

 ascertain the limit of pressure attainable with a screw, but the 

 same construction would serve as well for a pressure gauge in 

 connection with any form of compressor by removing the 

 screw B and inserting the connecting tube in its place. The 

 method presents the following advantages. The apparatus is 

 compact and, aside from the resistance boxes, etc., which are 

 to be found in every well ordered laboratory, is inexpensive. 

 It is free from leakage even at very high pressures. Observa- 

 tions can be quickly made and very easily reduced. The con- 

 stants, except a, can be easily determined. With reasonable 

 care the errors of measurement do not exceed one part in one 

 thousand. 



After a preliminary trial with a Bessemer steel screw, that 

 broke when a pressure of 3554 atmospheres had been obtained, 

 a bar of carefully annealed Stubb's steel was turned to fit the 

 thread in A as accurately as possible and then very evenly 

 tinned. Figure 3 exhibits the results of the final observations. 

 The abscissas indicate the angular positions of the top of the 

 screw, one division representing one whole turn, and the ordi- 

 nates show the corresponding pressures, the scale being 500 

 atmospheres per division. After five complete turns, when 

 the indicated pressure was 3600 atmospheres, the screw began 

 to twist and it finally broke after 7*7 turns when the maximum 



