ELECTEICAL RESISTANCE UNDER PRESSURE. 77 



These fragmentary runs have been given due weight in the final 

 results. 



No preHminary pressure seasoning of the wire was attempted. 

 This is usually unnecessary for soft metals, and in this case was 

 undesirable because of the effect of chemical action. The amount of 

 chemical action may be estimated from the amount of the change of 

 the zero after a run. At 0° the zero change was 5.6% of the total 

 pressure effect; at 25° 8.1%; at 50° 1.9%, and at 75° at 1000 kg. 

 4.7%. The smaller effects at 50° and 75° are because the Nujol mix- 

 ture was used at these temperatures to transmit pressure instead of 

 kerosene. Since the readings with increasing and decreasing pressure 

 were made at uniform time intervals, the mean zero should contain 

 little error from corrosion. Aside from the zero displacements, the 

 points at high pressures lay very regularly on smooth curves. At 0° 

 the greatest departure of any point from a smooth curve was 1.6%; 

 at 25° 1.3%, at 50° 0.8%, and at 75° 1.0%. 



The final results were obtained by first smoothing independently 

 the results at each temperature, and then smoothing the runs at each 

 temperature so as to give smooth temperature differences. The 

 maximum adjustment in this temperature smoothing was at 50° and 

 3000 kg., where an increase in the observed readings of 1.2% of the 

 total effect was necessary. 



The temperatiu-e coefficient at atmospheric pressure was obtained 

 from a coil of bare wire similar to that of the pressure measurements. 

 The details of the measurements were exactly the same as for lithium. 

 The two readings at 0° differed by 5.4% of the total temperature 

 effect. The relation between temperature and resistance can be 

 expressed by a second degree equation in the temperature. The 

 results at even temperature intervals are included in Table III. The 

 resistance of the solid at 100° (melting point 97.62°) may be extra- 

 polated from readings between 0° and 75° and gives as the average 

 temperature coefficient between 0° and 100° 0.005465. Northrup ^ 

 gives for the temperature coefficient of sodium in glass the value 

 0.0053, obtained by a linear extrapolation of values between 20° and 

 93.5°. Bernini ^ gives for the same temperature range (0° to 100°) 

 0.00428. As already remarked, there seem to be no previous values 

 on the unconstrained metal. 



The measurements on the resistance in the domain of both liquid 

 and solid above 97.6° were made with the sodiiun enclosed in a glass 

 capillary. The details were exactly the same as for potassium. In 

 point of time the measurements of potassium were made first, and the 



