ON STANDARDS FOE USE IN ELECTRICAL MEASUREMENTS 



437 



coefficients for the different parts were fonnd to be as follows : "028, "032, 

 •029, -036, 0-031, and 0-029 per cent, per degree. 



To examine the effect on the temperature- coeffi.cient of a variation in 

 the percentage of platinum and silver, the following alloys wei'e cast by 

 Mr. Hockin and drawn to wire. They were made with pure platinum, 

 black and precipitated silver. 



From the foi'egoing tables it would appear that a moderate variation 

 in the percentage-composition on either side of the normal proportions of 

 two parts silver, by weight, to one part platinum, produces less effect on 

 the temperature-coefficient than does a change in the diameter of the 

 wire. 



It also appears, from these and other experiments, that it is practically 

 impossible to ensure a uniform mixture of the metals, even when the 

 alloy is of the normal composition in the aggregate. 



Thus the wire from an unit coil of the B.A. pattern, made by 

 Dr. Matthiessen, or under his superintendence, was drawn down to a 

 diameter of 0'168 m.m. 



The whole wire had a temperature-coefficient of "0250%, but on 

 examining the wire in two approximately equal portions, the temperature 

 coefficients of the two halves were found to be ^0237 and '0266%. 



It is therefore evident that when the highest attainable accuracy is 

 acquired, as in the construction of standard coils, it is not sufficient to 

 depend upon the general temperature-coefficient of the alloy, but that a 

 detex'mination of the coefficient of the particular coil or instrument is 

 required. 



For less accurate work it would seem that Dr. Matthiessen's value for 

 the coefficient, viz., -031% per 1" C. should be reduced by from 5 to 

 10% for wires of large diameter, hy about 15% for wires of 0'25 m.m. 

 diameter, such as, in general, resistance-coils of from 100 to 1,000 ohms 

 are made of, and by from 20 to 25% for wires of the smallest gauge 

 usually drawn. 



Calibration of Wire. 



The calibration of the wire was effected thus — Six coils of the B.A. 

 pattern, viz., \a, \l, 2, 3, 5, and 8 ohms, whose resistance amounted, in 

 the aggregate, to 20 ohms, were arranged in a continued series by means of 

 mercury cups, and the ends of this series were connected by copper bars 



