582 BRIDGMAN. 



initial change. With a little practice a secondary change of such 

 magnitude may be found that temperature equilibrium is reached in 5 

 to 7 minutes. 



When the apparatus was in good running order and the thermo- 

 stat had reached equilibrium, a complete run on one substance at one 

 temperature could usually be made in about two hours. Including all 

 manipulations of adjusting temperature of the thermostat and wait- 

 ing for zero equilibrium, runs at two different temperatures could be 

 easily made on a single substance in a working day. 



General Character of Results. 



The effect of pressure on all the metals tried, with the exception of 

 antimony and bismvith, is to decrease the resistance. To a first 

 approximation, the relation between pressure and resistance is linear. 

 The average pressure coefficient is the datum which is of most immedi- 

 ate interest. The average coefficient up to 12000 kg. for any tempera- 

 ture was found by connecting the average zero with the average of the 

 two maximum points (which in many cases were nearer 13000 than 

 12000 kg.) by a straight line, calculating and plotting the deviations 

 of the observed points from this straight line, passing through these 

 deviation points a smooth curve (one is shown for lead in Figure 2), 

 and from this smooth deviation curve reading off the value of the 

 deviation at 12000, which may then be combined with the slope of the 

 straight line to give the average coefficient to 12000. The coefficient 

 found in this way is called the "observed" average coefficient. This 

 " observed " coefficient is a function of temperature. The " observed " 

 coefficient at each of the five temperatures was then plotted against 

 temperature, a smooth curve drawn through the five points, and the 

 value given by the smooth curve at any temperature taken as the 

 best value of the coefficient for that temperature. In the following the 

 smoothed average coefficient is given in tables as a function of tem- 

 perature, and the experimental points are given in diagrams from 

 which an estimate may be formed of the experimental accuracy. 



To a second approximation the relation between pressure and 

 resistance is not linear, but the initial rate of decrease of resistance 

 is in all cases greater than that at higher pressures. The departure 

 from linearity may vary with the substance from 0.8% to 5% of the 

 total change of resistance under 12000 kg. For some substances 



