Form of Resistance-Balance. 115 



time the coils are cooling ; so that the two readings are never 

 made under the same circumstances as regards temperature. 

 Beginning, then, with all four coils at the same temperature, 

 we take the difference between 3 and 4. To get them all at 

 the same temperature, water from the town mains is allowed 

 to circulate through the system for half an hour. At the end 

 of this time the difference of 3 and 4 is taken; and several read- 

 ings are taken at small intervals of time to see if the tempe- 

 ratures are constant. This being done, the temperature of 

 coil 4 is raised by the introduction of warm water until it is 

 about 15° above that of coil 3. It is best to raise the tempe- 

 rature about 20° above the other at first, and keep it there for 

 20 minutes, and then let it fall very slowly. In this way coil 

 and water cool together, and an equilibrium of temperature is 

 established between them. The difference between 3 and 4 is 

 again taken; and from these two readings we have, as seen 

 above, the mean variation-coefficient between the two tempe- 

 ratures. Another method, which would probably be a more 

 accurate one, for obtaining the mean coefficient of variation 

 between 0° 0. and 15° C. would be to wait until the tempe- 

 rature of the water in the town mains was about 15° C, and 

 then to keep three of the coils at that temperature, and to cool 

 the fourth by means of ice to zero. If then all four were kept 

 at 15° and the observations repeated, we should have the 

 means of finding the variation-coefficient of the fourth coil 

 between 0° and 15°. Prof. Chrystal, in his report, threw out 

 the suggestion that resistance-coils should have a thermoelec- 

 tric couple attached to them, one junction being buried in the 

 heart of the paraffin surrounding the wire, and the other out- 

 side. This has been tried in some coils recently made, and 

 proves a satisfactory method of ascertaining the equilibrium 

 of temperature between the wire and the water. 



Another source of error in the ordinary methods arises from 

 uncertain or variable resistances at the mercury-cups. It is 

 important that the copper legs of the coil-terminals should 

 press very firmly against the tops of the copper pins on which 

 the india-rubber-tube cups are fixed. To ensure this, the plan 

 adopted is to fasten on the coil-legs an ebonite clamp. Along 

 the edge of the wooden promontory, j h i (fig. 1), are put brass 

 pins m ; and by means of steel spiral springs fixed to these and 

 attached to the clamps the coil-legs are pressed down very 

 firmly (see fig. 3). The ends of the pins which carry the 

 india-rubber cups and the ends of the coil-legs being well amal- 

 gamated, we get, when they are thus firmly pressed in contact, 

 a very good joint, and one whose resistance is small and con- 

 stant. If the clamps are not used, then one leg may get lifted 



12 



