PRESSURE ON RESISTANCE OF METALS. 613 



bridge in the bath external to the cylinder. The coil on which the 

 pressure measurements were made showed essentially the same value, 

 0.00.3175, but the results were a trifle less regular. The abnormal 

 deviation of the resistance of palladium from linearity with the tem- 

 perature is to be noticed; the slope of the resistance-temperature 

 curve is greater at 0° than at 100°. This behavior is also shown by 

 platinum, but is the opposite of that of all the other metals previously 

 listed in this paper. This palladium cannot be very pure, because 

 its temperature coefficient is low. Kaye and Laby give 0.0037, and 

 Waidner and Burgess from the Bureau of Standards give .00333 to 

 .00337. The latter value is probably as good as that of Kaye and 

 Laby. 



For the initial pressure coefficient at 0° Beckman ^ gives O.O5219 

 against O.O5I9S3 found graphically from the deviation curves above. 

 Beckman does not give the temperature coefficient of his palladium. 



When the resistance-pressure cur\es are scaled to the same initial 

 resistance, the curves at higher temperatures are less steep. At first 

 the curvature becomes greater with increasing temperature but 

 passes through a maximum and becomes rapidly less. 



Platinum. This was the best Heraeus platinum, used in thermom- 

 etry. I am indebted for it to Professor H. N. Davis, who before 

 letting me use it had made a number of measurements on its tempera- 

 ture coefficient. The wire was wound bare on a bone core, was of 

 0.0038 inch diameter, and had an initial resistance at 0° of 21 ohms. 

 It would have been better if I had been able to obtain a longer piece, 

 but all the usual sources of supply were closed, and I was fortunate to 

 obtain this. It had been annealed to a red heat a number of times 

 by Professor Davis; after winding on the bone core it was still further 

 annealed by carrying back and forth between 20° and 130° a number 

 of times over an interval of four hours. It was seasoned for pressure 

 by one application of 12000 kg. at 0°; a decrease of resistance of 2% 

 of the total pressure effect was produced by the preliminary applica- 

 tion. It was soldered with ordinary soft solder to copper leads. 



The smoothed results are collected in Table XV, and the observed 

 values of mean coefficient and deviation from linearity are shown in 

 Figure 17. Because of the too low resistance the results are more 

 irregular than usual. There was an unexplained systematic discrep- 

 ancy, amounting to not more than 3%, at the maximum pressure at 0°, 

 25°, and 50°. Except for these irregularities at the maximum the 

 greatest departure of any single point from a smooth curve was 0.5% 

 of the pressure effect, and the average numerial departure was 0.14%. 



