ELECTRICAL RESISTANCE UNDER PRESSURE. 87 



values found were 1.664 and 1.56 at 62.5° and kg., 1.493 at 95.7° and 

 2260 kg., 1.550 at 132.2° and 5650 kg., and 1.550 at 167.0° and 10000 

 kg. The higher of the two values at atmospheric pressure was not 

 satisfactory because the capillary was not filled as well as usual, there 

 being obviously minute cavities between the surface of the metal and 

 the glass, and that result was accordingly discarded. The remaining 

 results show that the ratio is almost exactly constant, and in comput- 

 ing the tables the following values were used; 1.56 at kg. 1.555 at 

 2200 kg., 1.550 at 5430 kg., and 1.550 at 9710 kg. 



The value for the ratio of the resistance of liquid to that of solid at 

 atmospheric pressure may be compared with that of other observers. 

 Bernini '^ has found 1.392, and Northrup ^ 1.53. It is to be seen that 

 my value agrees much better with that of Northrup. This entire 

 c^uestion of the ratio of the resistance of the liquid to that of the solid 

 is still in a most unsatisfactory state experimentally, and results by 

 different observers disagree by much more than can be accounted for 

 by errors of measurement or by impurity of the materials. All values 

 with which I am acquainted, both for potassium and other metals as 

 well, have been obtained from measurements of the liquid in a glass 

 capillary, and from measurements of the solid in the same capillary 

 after freezing. The resistance of the solid is without doubt likely to 

 be largely in error because of strains and because of cavities formed 

 during freezing. Matthiesen,^ in his early work, found discrepancies 

 in measurements of the resistances of the solid of as much as 5% Avhich 

 he traced to this cause. What is more, after the solid has once been 

 formed, and is again melted to the liquid, there may be cavities be- 

 tween the surface of the liquid and the glass. Bernini has found large 

 discrepancies in the resistance of the liquid due to this cause. Further- 

 more, in most experiments, there has been a film of oil between the 

 surface of the metal and the glass; irregular capillary effects in this 

 film will introduce error. Much of the previous work should be 

 repeated with increased precautions. Measurements on the liquid 

 should be made with a capillary completely filled in vacuum, and the 

 liquid should not be allowed to freeze. The specific resistance of the 

 solid should be obtained from independent measurements of the bare 

 metal, preferably extruded to ensure complete freedom from cavities. 

 Measurements on both solid and liquid should be made over a wide 

 enough temperature range to allow unquestionable extrapolation to 

 the melting point. No measurements should be given much confi- 

 dence which show premature rounding of the corners of the melting 

 curve. It can now be stated with confidence that all such premature 



