in Terms of the Mechanical Equivalent of Heat. 5 



connecting wires. The whole was measured by connecting 

 the terminals of the circuit with a Jenkin-bridge and com- 

 paring with other coils, using a high-resistance Thomson- 

 galvanometer. The provisional standard was a 10-ohm coil, 

 A, made by Warden, Muirhead, and Clarke. From this coil 

 the resistance of a 100-ohm standard, B, was obtained by means 

 of a comparator, C, of ten coils, each nearly equal to A, each 

 coil of C being compared with A, and in series then com- 

 pared with B. Then A, B, and C were arranged to form a 

 bridge with D, a 1000-ohm standard, whose resistance was 

 thus fixed. E and F, two 1000-ohm coils of a resistance-box, 

 were then compared with D. Finally, a bridge was formed 

 with A, B, D-4-E+F, and B/, the secondary circuit, giving 

 R/ in terms of A. Elliott's coils were used in making the 

 adjustments, which were always very small, and the tempera- 

 tures were carefully observed. The result is 



B'=30012-4 at 19°-3 C. 



R/ consisted principally of the 30,000-ohm coil, and the varia- 

 tion of this only need be considered. Its temperature varied 

 from 19° to 24° when in use, the mean temperature being 

 22°*3 C. At this temperature R/ = 30052, which value was 

 used throughout. 



The length of the wire in the fixed coil was determined by 

 measuring with a steel tape the distance between two threads 

 fastened on the wire before it was placed on the circle. When 

 the wire was in position, the interval of a few centimetres 

 between the threads was measured. The tape had been com- 

 pared with standards. Care was taken to avoid difference of 

 tension in the two positions of the wire. The result is 



1= 264-49 centim. 



The quantity b (the excentricity of the needle) was esti- 

 mated by holding the tape horizontally over the top of the 

 circle, and reading the positions of the centre of the wire and 

 the galvanometer-fibre. For most experiments 6= 1*2 centim. 

 It was frequently remeasured, and a correction applied when 

 it varied. The method of measurement is not very accurate; 

 but an error of 10 per cent, in Z>, which could hardly occur, 

 would only involve an error of 1 part in 3000 in J. 



G / had been determined by Professor Rowland* by mea- 

 surement during the construction of the coil, and also by 

 comparison with another coil. The values are 1832*24 by 



* Silliman's American Journal, 1878. 



