PRESSURE ON RESISTANCE OF METALS. 597 



spring clip. The error in using the ch'p is introduced hy the heavy 

 coating of oxide, which forms rapidly on magnesium in the air, and has 

 very high insulating qualities. With this clip a series of regular 

 results were obtained between 12000 and 6000 kg., showing the same 

 linear relation between pressure and resistance as the other run, but 

 no other points of this run were at all good. It did not seem worth 

 while to try for points at other temperatures than 0°. The pressure 

 coefficient at 0° of this sample of magnesium may be taken to be 

 — O.O5.55 within perhaps 1 or 2%. No attempt was made to find the 

 deviations from linearity. 



There are no previous measurements of the pressure coefficient over 

 any range. For the temperature coefficient I found .00390 between 

 0° and 20°. This may be compared with 0.00381 between 0° and 100° 

 by Dewar and Fleming. ^^ The material was not very impure, evi- 

 dently. 



Aluminum. This is one of most difficult of metals to get entirely 

 pure. I obtained some considerably purer than that on which meas- 

 urements are usually made through the courtesy of the Aluminum 

 Company of America. Their analysis was Fe 0.23%, Si 0.24% 

 Cu 0.06%, Al 99.47%. It was provided by them in the shape of ^ 

 inch rod. I extruded it hot from this size to 0.06 inch diameter and 

 etched the surface to remove iron. It was then drawn down to 0.005 

 inch through diamond dies and single covered with silk insulation by 

 the New England Electrical Works. About 30 ft. of it was wound 

 on itself into a coreless toroid and seasoned by carrying several times 

 back and forth between 0° and 120° over an interval of 5 hours. Its 

 initial resistance at 0° was 18.7 ohms. 



Great trouble was found in making connections to this fine wire. 

 Commercial aluminum solder was tried without success, the fine wire 

 being completely alloyed through and eaten off at the temperature 

 necessary to make a good connection. Finally connections were made 

 with a spring clip but with results not entirely satisfactory. Consist- 

 ent irregularities of as much as 4% of the total effect were found at 

 the last three points with decreasing pressure at all temperatures, due 

 probably to some slight slipping of the clip. The zero, however, was 

 recovered satisfactorily; the permanent changes of zero were 1.4, 0.0, 

 0.6, 0.2, and 0.7% of the total effect. Except for the points at which 

 consistent departures were found the greatest departure of any point 

 from a smooth curve was 0.5% of the total effect. With a perfectly 

 satisfactory method of making connections it might possibly pay to re- 

 peat these measurements searching for a polymorphic transition. The 



