430 REPOKT — 1881. 



showed no deflection. In this way the third line on the recording strip 

 was displaced to one side whenever the speed was such as to cause the 

 electromotive force of the coil and that due to the thermo-electric pile 

 accurately to balance each other, and thus the parts which were to be 

 measured of the other two lines were indicated. A second observer noted 

 the reading of the tangent-galvanometer when the zero galvanometer 

 was undeflected., and thus determined the angle 0. 



It is evident, from the formula given above, that the product T tan d 

 should be constant in all experiments in which the wire whose resistance 

 was to be determined was the same. The amount of agreement in the 

 value of this product in different experiments therefore affords a criterion 

 of the consistency of the results with each other. The results obtained 

 in two series of experiments were as follows : — 



T tail 9 T tan 9 



(Series I.) (Series II.) 

 001291 • 001192 



•01296 -01196 



•01309 -01194 



•01312 -01196 



-01309 -01189 



•01283 ^01192 



•01298 ^01193 



•01306 -01194 



•01295 — 



•01306 — 



-01302 — 



•01294 — 



■01310 — 



It will be seen that the second set of values agree better together than 

 the first set. This is probably chiefly due to greater practice in observing, 

 and to the adoption of an artifice whereby the speed of the gas engine 

 employed to drive the coil was kept more constant. In calculating the 

 final result from each set, weight was given to each separate observation 

 in proportion to the square of the number of revolutions of the coil over 

 which it extended ; for it was assumed that the accuracy with which the 

 speed of the coil was determined was proportional to the number of 

 revolutions included in the record ; moreover, the number of galvanometer- 

 readings obtained in each experiment was proportional also to the number 

 of revolutions, and hence it was assumed that the accuracy with which 

 the product was determined was proportional to the square of the number 

 of revolutions. The weighted means thus calculated are, for the 

 First series . 0'013017 

 Second series . •011932 



Calling p■^ and p2 ^'be resistances measured in the two series of experi- 

 ments respectively, these results give 



and p, = ^|f~~ X 10« = 80-089 x 10^ 



The wires measured belonged to a set of german-silver resistance- coils, 

 which were very carefully adjusted by my assistant, Mr. W. Grant, by 

 comparison with a ' B. A. -unit ' coil issued by the former Committee. 

 The nominal values were 73 and 80 ohms in the two sets of experiments 

 respectively. Applying a not very certain correction for the difference 



