402 Prof. Ayr ton and Mr. Medley on 



are the main terminals of the lamp-stand, corresponding with 

 M and N in fig. 7. From M the current, after passing 

 through the resistances e and d, for the two groups run at 

 lower pressures, went through the lamps and reached the 

 terminal N by means of the copper springs a, b, and c. 



When, however, a group of lamps was turned into position 

 for photometric measurements its copper springs c (J and K 

 in fig. Id) were separated by a piece of brass L backed with 

 insulation (R in fig. 7a), and so the current of the group 

 was diverted through a thick wide strip of manganin, this 

 strip being shunted with an Ayrton and Mather d'Arsonval 

 galvanometer, A, in series with a resistance R, as seen in 



This combination of galvanometer, resistance R, and man- 

 ganin strip was calibrated by direct comparison with a 

 Kelvin balance, and the resistance R was so adjusted that 

 one ampere produced a deflexion of 500 divisions, each of 

 about a millimetre in length. Variations of current during 

 the experiments could be read to 50V0 °f an ampere if 

 required, and the current was known accurately in Board of 

 Trade amperes to yxnro °^ an ampere. 



The results of frequent comparisons between an ampere 

 as read off on the direct-reading transparent scale and an 

 ampere as measured with the Kelvin balance, which was also 

 screwed down permanently in position, never showed a dif- 

 ference as large as 1 part in 1500 between September 1893 

 and August 1894. 



The positive and negative terminals of each group of lamps 

 were connected respectively with two squares of brass a 3 a 

 (fig. 7) let into the slate bed on which the group was 

 mounted. Two brass springs, Gr, H, permanently connected 

 with an Ayrton and Mather reflecting electrostatic voltmeter, 

 the motion of the needle of which was damped by moving in 

 a magnetic field, pressed on the brass squares a, a of which- 

 ever group of lamps was brought to the front of the stand. 



The zero of the voltmeter was so arranged that the tran- 

 sparent scale, which was direct-reading, was nearly uniformly 

 divided from 50 to 120 volts, each division of about 1*5 milli- 

 metre representing 0*2 volt, so that it was easy to read the 

 deflexion to within 2 parts in 10,000. 



The electrostatic voltmeter was frequently calibrated, and 

 the maximum change in sensibility between July 1893 and 

 August 1894 did not exceed 0*2 per cent. This change, 

 although small, is fully accounted for by the variation in the 

 zero, since the instrument did not possess the delicate zero 

 adjustment of the later specimens of this type of voltmeter. 



