A NEW CURRENT WEIGHER, ETC. 505 



of the various parts has been measured on several occasions, and the lowest measured 

 resistance between any two adjacent coils is 2000 megohms, and between the balance 

 circuit and earth it is 1000 megohms. The applied pressure in the former measure- 

 ments was, in general, 40 volts, and for the latter 200 volts. When the coils are 

 arranged in two groups, so that each group consists of one coil of each pair, the 

 insulation resistance is 1500 megohms. The first measurement was made in March, 

 1905, and the last in April, 1907. 



SECTION 8. ADVANTAGES OF DUPLICATING THE COILS. 



As previously mentioned, there is a set of coils at each end of the balance. Several 

 advantages are gained by this arrangement. In the first place, the force to be 

 measured is doubled by using the two sets of coils, and the accuracy of the measure- 

 ment is therefore increased. A much greater advantage, however, arises from the 

 symmetry thus obtained, for mechanical disturbing causes will, on the whole, tend 

 to be neutralised. 



One of the principal disturbances arises from convection currents produced by the 

 heat generated in the coils, and in the flexible connections to and from the suspended 

 systems. Another is the change of buoyancy due to change of temperature of the 

 air in which the suspended coils hang. Both these produce a fairly rapid drift of the 

 rest-point of the balance when a single set of coils is used, but when both sets are 

 employed the steadiness of the balance is greatly improved. The extent of this 

 improvement will be seen on reference to fig. 20, which shows four pairs of curves 

 taken to test this matter. During all these tests the adjacent helices on each cylinder 

 were connected up, so that the current (if any) flowed in opposite directions in 

 adjoining wires, thereby making the windings inoperative, and obviating the necessity 

 of keeping the current very constant. 



Several sets of about 12 readings of the swings of the balance were taken under 

 each of the following conditions respectively : 



(a) No current through either set of coils. 



(6) Normal current through both sets of coils. 



(c) ,, left-hand set of coils. 



(d) right-hand set of coils. 



The rest-points were calculated from each group of three successive readings 

 throughout a set, and the values tabulated, thus giving the rest-points for each half 

 period. From the several sets of observations taken under each of the conditions (a), 

 (b), (c), (d) respectively, those showing the least and greatest drifts were plotted, 

 the former being shown in full lines and the latter dotted in tig. 20. The points thus 

 obtained were joined by straight lines, and no attempt made to smooth out 

 irregularities. In this figure the middle of the balance scale is denoted by 100 ; one 

 division of the scale is about -^th of a millimetre (actually 0'095 inilliin.), and as this 



VOL. cc-vn. A. 3 T 



