PRESSURE ON RESISTANCE OF METALS. 609 



fore, expect the results to be more irregular. The greatest departure 

 of any point from the smooth curve is 0.8% of the total pressure effect 

 and the numerical average is 0.35%. The deviation from linearity is 

 symmetrical and parabolic within the limits of accuracy; there is no 

 need of exhibiting the deviations graphically. 



The mean temperature coefficient found above is 0.00365. Kaye 

 and Laby give 0.0033. It is probable that this value is quoted from 

 Reichardt^^, who gives 0.00326 for cobalt 99.8% pure. There seem 

 to be no measurements of pressure coefficient for comparison. 



A distinctive feature of the behavior of cobalt is the abnormally 

 low pressure coefficient. When the resistance-pressure curves are 

 scaled to the same initial resistance, the steepness is less by an 

 unusual amount at the higher temperatures, as is also the curvature. 



Iron. This material was American Ingot Iron, obtained from the 

 American Rolling Mill Co. of Middletown, Ohio. It is of exceptional 

 purity, showing less than 0.03% total impurity. Under microscopic 

 analysis, done under the direction of Professor Sauveur, it appeared 

 to consist entirely of ferrite. It is much purer than most samples 

 which have been prepared in small amounts by the most refined 

 methods in the laboratory. It was not furnished by the manufacturers 

 in the form of wire, but this particular sample was provided in a sheet 

 jg inch thick. From this a square wire was cut with the shears, 

 the corners rounded with the file, and it was drawn to 0.04 inch through 

 steel dies. It was then heavily etched and drawn to 0.005 inch through 

 diamond dies and wound with single silk insulation by the New Eng- 

 land Electrical Works. It was annealed by heating to redness after 

 the final drawing. For the pressure measurements it was wound in- 

 to a coreless toroid of 70 ohms resistance at 0°. Connections were 

 made with silver solder. It was seasoned by heating repeatedly to 

 140°, and by several applications of 12000 kg. 



The smoothed values are shown in Table XIII and the experimental 

 values for mean coefficient and departure from linearity in P'igure 15. 

 The maximum zero drift was 0.03% of the total pressure effect. This 

 was one of the earliest substances investigated, and pressure was 

 transmitted at all temperatures by kerosene. The results are some- 

 what more irregular than usual. Except for four bad points, the 

 maximum departure of any individual reading from a smooth curve 

 was 0.54% of the total pressure effect, and the numerical average was 

 0.12%. Within the limits of error the deviation from linearity was 



15 G. Reichardt, Ann. Phys. 6, 832-855 (1901). 



