On a Resistance Comparator. 673 



are connected by a short copper strap. When, as in the 

 potentiometer-coils tor the measurement of which this com- 

 parator was designed, there is an appreciable connecting 

 resistance between the coils to be compared, the two-point 

 bridge method is used; the strap is then removed and an 

 adjustable resistance inserted. The ends b g of the comparison 

 resistances are connected by copper straps to the cups of a 

 mercury commutator, and through that to a pair of large 

 binding screws that project outside the board carrying the 

 coils. 



When the galvanometer switch is set on d or e each arm 

 is 100 ohms ; as it is moved round the dial the resistance is 

 altered by steps of ^ part. The galvanometer deflexions are 

 taken for the two positions nearest balance and interpolation 

 to ^ calculated. In this way I consider that an accuracy of 

 one part in 100,000 is attainable. 



To calibrate the comparator the two arms are adjusted to 

 be as nearly equal as possible, and then compared by the use 

 of a pair of nearly equal resistances measured in the usual 

 way. Next the resistances from h to d and from e to g are 

 separately compared with a standard 100 w , and c to d and e to 

 / with a standard ohm. Finally, a good box is put in parallel 

 with cd and ef in turn and the ratio corresponding to each 

 stud determined. This need only be done to ^ part to ob- 

 tain logout) m f ne nua l results. 



The method of interpolating by the galvanometer has, I 

 think, been unduly neglected. Even when one tries to carry 

 out a strictly null experiment, one is obliged, on account of 

 thermoelectric effects, and so on, to observe accurately minute 

 movements of the galvanometer needle. It therefore does 

 not introduce any new difficulty to read the galvanometer- 

 scale exactly each time, and it may be made, as in this com- 

 parator, to avoid reading the position of a sliding contact on 

 a scale. I think therefore that there is gain of accuracy as 

 well as of convenience in using the interpolation method. 



The galvanometric arrangements I have adopted for thi* 

 purpose are as follows: — A plane mirror, 15 mm. diam., is 

 attached to the coil of the (D'Arsonval) galvanometer. A 

 telescope of 25 millim. aperture is placed some 600 millim. in 

 front, and, mounted on the telescope-tube, one of Zeiss's trans- 

 parent glass scales, backed by ground-glass and illuminated 

 by a small incandescent lamp behind. The mirror ami tele- 

 scope were made by Mr. Hilger : the scale is 100 millim. 

 long, and the graduations are exceedingly fine. Altogether 

 the definition is so good that one can read to .,' millim., i. e. to .^ 



