ELECTRICAL RESISTANCE UNDER PRESSURE. 125 



numerically than for the other piece, and the departure from linearity 

 was much greater, suggesting strongly that at a high enough pressure 

 the resistance may pass through a maximum. The following increases 

 of resistance were found at 2000, 4000, 6000, 8000, and 10000 kg. 

 respectively; namely, 0.69%, 1.25%, 1.62%, 1.79%, and 1.86%. 

 The temperature coefficient of resistance of this specimen was not 

 measured. 



Although the results obtained above are not of much accuracy and 

 are not reproducible, two interesting facts stand out; the opposite 

 signs of the pressure coefficient for carbon in the amorphous and 

 graphitic states, and the large departure of the effect from linearity 

 with graphite, indicating a maximum. No substance has yet been 

 found in which a maximum or minimum of resistance has been actually 

 reached at high pressures. 



Silicon. It is well known that technical means are as far from 

 perfect for producing a pure and reproducible silicon as they are for 

 carbon. For instance, the temperature coefficient of resistance varies 

 in sign with different pieces of apparently the same manufacture. 

 In view of this situation it was worth while to make only a few pres- 

 sure measurements in order to establish the general nature of the 

 effects. Two specimens were used, both provided by the General 

 Electric Co. 



The first sample had not been manufactured by them, but had 

 been obtained from the Carbonmdum Co. It was in the form of a 

 cylinder about 5 mm. in diameter and 8 cm. long. Connections were 

 made with spring clips, and measurements made by the potentiometer 

 method, as usual. The effects were very irregular; I satisfied myself 

 that the irregularities were inherent in the material itself. There 

 were large seasoning effects on the first application of pressure, there 

 were always permanent changes of zero after a run, and there were 

 differences between the readings with increasing and decreasing pres- 

 sure in a direction the reverse of hysteresis. Two series of runs were 

 made, at 0° and 52°. The resistance decreases under pressure, as is 

 normal. At 0° the total decrease under 12000 kg. was 14.0%, and at 

 52° 15.8%. The effect is not linear with pressure, but the coefficient 

 becomes less at the higher pressure, as is normal. The average 

 temperature coefficient of this sample between 0° and 52° was 0.000117. 



The second sample was also furnished originally by the Carborun- 

 dum Co., but it had been partially purified by the General Electric Co. 

 by melting in vacuum. The purification was by no means complete, 

 for it was possible to see with the naked eye small slag-like inclusions 



