72 



BRIDGMAN. 



pressure given in the Table was extrapolated from the three observed 



.1, • 1, ^'^ • 



points on the assumption that — is constant. 



dp- 



The most important result of these melting observations is that the 

 melting curve is normal, pressure and temperature rising together. 

 This means that the solid contracts on freezing, as is normal. The 

 positive pressure coefficient of resistance of the solid therefore need 

 not be due to the peculiar mechanism that it is in the case of bismuth. 



The changes of volume on melting are very small, and because of 

 the rounding of the corners, very hard to determine accurately. The 

 error was too great to establish any regular variation of the change 

 along the melting curve. The best mean value for the fractional 

 increase of volume on melting is 0.006. 



Pressure, Kg. ,. Cm." X 10' 

 Lithium 



Figure 2. The melting curve of lithium. 



The effect of pressure on the resistance of the liquid was measured 

 on the same sample as that of the melting curve, the sample from 

 Merck having been used up by the chemical analysis. Considerable 

 difficulty was mot in de^^sing a suitable method. It is well known 

 that melted lithium attacks glass, and this I A'erified by my own 

 experience. Pre\'ious experimenters have been able to measure the 

 resistance of the liquid in a glass capillary by protecting the interior 

 surface of the capillary with a film of oil. This introduced error 

 because of the space occupied by the film of oil, and is not adapted to 

 use under pressure, the oil film being penetrated after a few compres- 

 sions. The method finally adopted was to surround the lithium with 



