PRESSURE ON RESISTANCE OF METALS. 579 



1.7 cm. diameter and 9 cm. long, and the wire was wound loosely in 

 the groove. 



Another mechanical efPect apart from that offered by the constraints 

 is due to the viscosity of the transmitting medium. This was of course 

 particularly prominent at low temperatures. It may be almost 

 entirely avoided by using petroleum ether to transmit pressure at 0° 

 and 25°. At higher temperatures, pure kerosene may be used without 

 sensible error. 



When these various precautions have been taken to avoid extra- 

 neous mechanical and temperature effects, results may be obtained of a 

 constancy and regularity much greater than I had anticipated. In 

 the majority of cases pressure could be applied to 12000 kg. and 

 removed with a change of zero of less than 0.3% of the total change, 

 or a constancy of the total resistance of 0.01%, and in many cases the 

 change was imperceptible. This change was not the effect of pressure 

 alone, but was the sum of all effects, including temperature drift in 

 the thermostat and changes in the bridge due to changes in room tem- 

 perature. Nevertheless, 0.3% of the change might be 15 or 20 times 

 the sensitiveness of the measurements, and in all such cases a correc- 

 tion for the zero drift was applied proportional to the time. 



The wires were attached to the terminals of the insulating plug 

 with silver solder in most cases where the melting point was high 

 enough; the softer metals were attached usually with soft solder. 

 In a few cases other methods of contacting were employed, which will 

 be described in detail later. In all cases in which metallic connection 

 can be made by fusing, there need be no trouble whatever at the con- 

 tacts. 



Other minor corrections are for the effect of pressure on the leads 

 of the insulating plug (only 0.2 mm. of bridge wire at the maximum) 

 and a temperature correction for the leads, which was determined 

 experimentally, and the correction for lack of uniformity of the bridge 

 wire. 



The purity of the metals used is a matter of great importance. The 

 harder metals were drawn from sizes below 0.04 inch through diamond 

 dies. Before passing to the diamond dies the trace of iron that might 

 have been rubbed in from the larger steel dies was removed by etching 

 off with acid at least 8% of the diameter. Softer metals were extruded ■ 

 at one operation through a steel die. Special tests with the very 

 delicate potassium thiocyanate method showed no perceptible iron 

 introduced during the extrusion. Chemical analyses are given for all 

 those metals for which I could obtain it. As giving the best indication 



