98 BRIDGMAN. 



obtain the analysis from the paper of Glascock, because the products 

 of his separate electrolyses were not always of the same purity, and 

 the different batches were indiscriminately mixed in the material as 

 supplied to me. The least pure of any of the specimens of Glascock 

 had about 2% impurity, and the best about 0.15%. 



Pressure measurements were made in the regular way with the 

 potentiometer. The wire was used bare, and the connections were 

 made with spring clips. The resistance of the contacts increased 

 during the runs, and sometimes became troublesomely large; it could 

 then be reduced by passing a high tension current from a small magneto 

 through the contacts. The wire was seasoned by a preliminary appli- 

 cation of 3000 kg. at room temperature; there was no perceptible 

 change of zero after this application. Three runs were made; at 0°, 

 50.5°, and 97°. At the highest temperature the zero of pressure was 

 taken as 500 kg. in order to prevent chemical action, and the value 

 at atmospheric pressure obtained by extrapolation. Considering the 

 chemical activity of this material the readings showed a gratifying 

 regularity. At 0° there was a permanent change of zero of 1.8% of 

 the total effect, at 50° the change was 0.3%, and at 97° 2.1%. The 

 maximum departure of any of the other points from the smooth curve 

 was 1.6% at 0°, 1.2% at 50°, and except for one bad point, 1.5% at 

 97°. The observed resistances were smoothed and a table con- 

 structed for the resistance at regular intervals of pressure and tempera- 

 ture in the regular way. 



The results are shown in Table VIII and Figure 7. The resistance 

 increases under pressure, the same as for calcium. The increase is 

 furthermore \'ery large; it is five times as large as that of calcium, and 

 three times that of bismuth, and is the largest positive coefficient 

 yet found. The behavior is in other respects like that of other metals 

 with a positive coefficient. When the resistance is plotted against 

 pressure, the curve is concave upwards, the change becoming more 

 rapid at the higher pressures. The coefficient becomes markedly 

 smaller at the higher temperatures. The instantaneous coefficient 

 becomes smaller at the higher pressures; this was not the case for 

 bismuth. One may be puzzled at first by the Table which shows a 

 smaller instantaneous coefficient at bothO kg. and 12000 kg. than the 

 average coefficient between and 12000. The reason for this is that 

 the instantaneous coefficient is calculated in terms of the resistance 

 at the pressure in question, which becomes rapidly greater at the 

 higher pressures, whereas the average coefficient is calculated in terms 

 of the initial resistance at kg. The resistance shows a regular drift 



