PRACTICAL STANDARDS FOR ELECTRICAL MEASUREMENTS. 37 



and in somewhat wider tubes of specially thin glass, through which the 

 four leads are hermetically sealed. The heads of all these thermometers 

 are of the design used by Chappuis and Harker, the contacts to the solid 

 ends of the copper flexibles being made by fusible metal cups. "With 

 reasonable care these contacts prove very satisfactory, both as I'egards the 

 constancy of their resistance and their mechanical strength. 



In the construction of all these thermometers special care was devoted 

 to adjusting their fundamental intei'vals to be very close to their nominal 

 values, and after completing this adjustment all were subjected to repeated 

 annealing in air at a bright-red heat, thermometers Nos. 3 and 4 being 

 temporarily placed in porcelain tubes for the purpose. 



The remaining four constructed last summer, and one of later date, all 

 of 1 ohm fundamental interval, have had their constants determined from 

 time to time during the year. One of them — B. A.j — was selected as a repre- 

 sentative platinum thermometer for use in an investigation made to deter- 

 mine the relation between the platinum scale and that of the gas thermo- 

 meter of the National Physical Laboratory at temperatures up to 1000° C 

 During the time occupied by two sets of experiments with this instrument, 

 extending over about three months in all, its constants altered by an 

 amount only just greater than their probable error, showing that it is quite 

 possible to use properly constructed platinum thermometers up to tem- 

 peratures slightly over 1000° C. for long periods without fear of serious 

 changes. 



The summary of the life history of the different thermometers is 

 given in Appendix III. The chief fact apparent is that there seems to 

 be a small but real difference between the 8 of thermometers 1 and 3 on 

 the one hand, and 2, 4, and 7 on the other, the maximnm divergence 

 being about '02. 



Prolonged electrical heating in air of the wire of one of the thermo- 

 meters was not found to sensibly change the value of the S. The cause 

 of the small differences found is not obvious, and further investigation is 

 being made on this point. 



A change in S from l-.'oO to 1-.51 would make at the sulphur-point a 

 difference of 0°-153 C, and at 1000° C. one of 0°-9. 



The question of the resistance of copper has been raised lately by the 

 work of one of the sub- Committees of the Engineering Standards Com- 

 mittee. For commercial purposes the resistance of copper is dehned at a 

 temperature of 60° Fahr. (15°-55 C). A table in Appendix lY. gives 

 the values that have been found by various experimenters. 



It is clear that copper is now prepared of a higher degree of purity 

 than in the time of Matthiessen. Taking the mean of the figures 

 in the table for modern electrolytic copper, we have as the value of the 

 resistance of 1 metre of copper wire weighing 1 gramme at 15° '55 C. the 

 value O'H'SSj ohm, but the figures of which this is a mean range from 

 1475 to 1492. The value found by Matthiessen, as deduced from his paper 

 in the ' Phil. Trans.' for 1860, is 0-1500 ohm. Thus the conductivity of 

 modern pure electrolytic copper is 1 per cent, better than Mattliiessen's. 



The Committee on copper conductors, which investigated the question 

 in 1899, adopted the number 0*1508 ohm as the resistance of a metre- 

 gramme of commercial annealed high-conductivity copper. This figure 

 has been accepted by the Engineering Standards Committee. 



Mr. H. A. Taylor has recently placed in the hands of the Secretary 



