ELECTRICAL RESISTANCE UNDER PRESSURE. 89 



magnesium and because of the very high resistance of the fihn of oxide 

 on the surface. With a potentiometer method of measuring resistance, 

 however, this was no longer a difficulty, and accordingly the attempt 

 was made to get more accurate Aalues. 



INIeasurements were made with two samples of magnesium. The 

 first was from the same piece as that on which measurements of resist- 

 ance and thermal e.m.f. have been already published, and was a 

 contiguous length from the same spool as the e.m.f. sample. It was 

 originally obtained from Eimer and Amend, of commercial ciuality. 

 The method of manufacture of magnesium, however, is such that 

 impurities are not likely to get into it, and it is a matter of experience 

 that commercial magnesium is of higher absolute purity than most 

 commercial metals. The second sample I owe to the kindness of 

 Dr. INIacKay of the Research Laboratory of the General Electric Co. 

 It had been especially purified by him by distillation in vacuum. This 

 was extruded hot to wire of about 0.020 inches diameter, the same in 

 dimensions as the other specimen. Both specimens were mounted 

 in the same way for the measurements, by winding bare on a bone core. 

 Contact was made ■v\dth spring clips. The resistance at. the contacts 

 was so high that error might be introduced because of fluctuations of 

 the potentiometer current, unless the precaution were taken to 

 brighten the wire with sandpaper immediately under the clips just 

 before assembling the apparatus. With this precaution no trouble 

 was experienced from contact resistance. 



The measurements on the first sample of magnesium were made just 

 after the apparatus had been constructed, and before all points in the 

 best handling of it had been settled, so that there were a number of 

 incomplete runs. In all, there were ten runs on this sample, five of 

 them complete. The incomplete runs were given due weight in the 

 final results. The maximum deviation of the individual readings 

 from regularity was of the order of 1%. It will not pay to reproduce 

 the results in detail, because this sample was presumably less pure 

 than the second, and the results are somewhat different. 



Three runs were made on the second and purer sample, at 0°, 51° and 

 95°, after three preliminary applications of 3000 kg. at room tempera- 

 ture to season. The results were smoothed and a Table constructed 

 in the regular way. The readings ran regularly. At 0° the zero dis- 

 placement after the run was 1.6% of the total effect, at 51° it was 1.2%,. 

 and at 95° 0.85%. These displacements are also essentially the same 

 as the maximum departures from a smooth curve of any of the other 

 observed points. The results are exhibited in Table VI and Figure 6^ 



