206 
Proceedings of the Royal Society of Edinburgh. [Sess. 
The current in the solenoid which establishes the transverse field is too far 
distant from the nickel to have any direct temperature effect upon it. 
In the first and second groups of five measurements, the current in the 
anchor-ring coil lasts only for a fraction of the whole time, producing an 
increase of resistance represented by 25*3 ( = 48*8 — 23*5) in the first group 
and by 27*7 in the second. In the third group, however, the longitudinal 
field current is flowing the whole time ; and the corresponding change of 
resistance is represented b}^ the number 73. These differences are all in 
the same direction, increasing positive or decreasing negative. Roughly 
speaking, the heating current in the first and second groups of experiment 
acts in a periodic manner for two-fifths of the whole time of the five 
measurements, whereas in the third experiment the current flows for the 
whole time. Now, two-fifths of 73 is 29*2, which is as close to 27*7, or even 
25*3, as we could reasonably expect. 
Finally, in the fourth group the differences are positive and diminishing, 
indicating therefore a decreasing resistance by an amount represented by 
the number 35*7. But this is quite in line with the general explanation ; 
for now there is no heating current in the anchor-ring coil, which begins to 
cool down towards its original temperature at about half the rate of its 
previous rate of heating. It is no doubt this heating effect which makes 
the experimental errors somewhat considerable in a few instances. 
The final reduction in the notebook is to the resistance change c£N /N, 
which is the ratio of the sum of the two final differences to the calibration 
number already obtained. This particular way of making the reduction 
is necessary because the first difference 30*1 in the calibration experi- 
ment corresponds with half the first differences in the other experiments, 
since these are obtained by the reversal and not merely by the making or 
breaking of the steady current through the Wheatstone bridge. Hence, to 
the sum of the four values of the calibration difference used in making the 
reduction will correspond the sum of the two final values of the deflections 
due to the magnetic change of resistance. The method is of course simply 
a convenient way of comparing means. 
Each change of resistance is expressed in terms of 100,000 ohms, and 
each final value is deduced from fifteen individual readings on the galvano- 
meter scale. The four groups of observations of the kind indicated above 
may be regarded as forming one complete experiment for a given combina- 
tion of longitudinal and transverse field. The four changes of resistance 
measured are produced by the longitudinal field ( h ) acting cyclically alone, 
by the same field acting cyclically when the transverse field ( t ) is main- 
tained, by the transverse field (t) acting cyclically alone, and by the same 
