AND RESISTANCE OF NICKEL WIRE AT HIGH TEMPERATURES. 43 
conditions, taking the largest value of the ratios dD/Dd\ and dD/DdL, namely, 
89/572, we find 
| 89 
Dai es od(M + MA ssa00) 
Dae BL ( Eyal =a 
showing at once that the second terms in the brackets are negligible, and that equation 
(5) still holds. 
Finally, consider the most extreme case of all, in which the applied field was so great 
as to produce a change dl, which made the spot of light travel from end to end of the 
scale. It was necessary in this case to disturb the balance on the bridge, so that the 
initial current 7 produced a deflection 3°5 times that due to the imposed change dh. 
That is to say, since the change dA meant an alteration of 0°002655 in the value of Lu 
or MA in the experiment corresponding to equation (2), the initial value of z in this 
extreme case of the experiment corresponding to (3) was such as to make 
Lu—MA = 3°5 x ‘002655 Lu = 0093 Lu 
and 
MX = -9907 Lye. 
Thus equations (4) become 
Ddi = - eMdr ] 
L =| } 
Dei =e epdL( 1 - 00935 
This extreme case occurred only at the ordinary temperature of 15° C. Putting in the 
corresponding values of L, D, and dD/dL, and taking the ratio, we find 
di_ Maar 1 
3 uw SL 1—-0093 x 3:12 x 0158 
ee iG 1 pes ae 
oot i ST 
Hence equation (5) is in error by 0°5 per cent.; and it will be noticed that this is due 
mainly to the factor by which we pass from the ratio M/u to L/A. The effect of the 
second terms in the brackets of equations (4) is in all cases negligible. Hence in every 
case we may write 
di_ Maa 
ae Ol 
and in the great majority of cases use the more convenient form (5). 
3. REDUCTION OF THE OBSERVATIONS.—The results embodied in Table A were 
plotted on a large scale, the change of resistance dL being plotted in terms of the field. 
In any one series of experiments the temperature varied a little throughout; but it was 
easy to apply slight corrections by graphical interpolation so as to obtain a series of 
isothermal curves. From these curves the values of dL were read off for the fields 
2,4, 6, ete., up to 34, and were then divided by the appropriate value Ly of the re- 
sistance of the nickel wire included in the coil. This was assumed to have the same 
