VALUE OF THE BRITISH ASSOCIATION' UNIT OF RESISTANCE. 
243 
taken as the temperatures at the time of the experiment, and the means of the values 
of x y , x y'. 
The alteration in temperature during the experiment was only about - 1° or - 2° 0. 
The changes in the values of x y, See., during the observation were produced by the 
variation of the temperature of the secondary coil, which being copper has a high 
temperature coefficient. The alteration in the value of x was rarely as much as 300 
bridge divisions, and since this change went on nearly uniformly during the experi¬ 
ment, the mean of the two values at the beginning and end will be very accurately the 
true value. 
In reducing the scale readings to give the throws and deflections, we had to 
remember that owing to the variation in the direction of the earth’s magnetic force, 
there was a continual, change going on in the resting point of the needle. 
In general, this change was only ‘3 or - 4 millim. during the time occupied by a set 
of observations, sometimes it amounted to 1 millim. or rather over, and on one or two 
occasions during magnetic storms the changes were so violent and sudden that we had 
to cease work entirely. 
For determining the value of the throw the following method enables us to eliminate 
the effect of this change in zero. 
Take the mean of the two throw readings to the right, then the mean of the two to 
the left, and the mean of the resting point readings, then the differences between the 
throw readings and the resting point, or zero readings, will give the throws right 
and left respectively, corrected for change in resting point. The difference between the 
throw readings will give the value of p x -\-p<i directly ; since, however, the throws right 
and left ought to be the same if the adjustments are correct—it forms a test of the 
accuracy of the measurements to calculate p x ar| d p 2 separately. 
For the permanent deflection, however, in which only one observation was made on 
either side of the zero, the same method is not applicable. The four observations we 
have to consider are : zero reading, deflection to right, deflection to left, zero reading. 
Suppose the zero is moving from right to left, then, if we take the mean of the zeros 
and consider the differences between it and the deflection reading as the deflections 
right and left, in each case our deflection will be too great, while, if the zero be 
moving in the other direction, the deflection obtained will be too small. 
To obviate the difficulty we assumed that the interval of time between each two con¬ 
secutive observations was the same, and that the change in zero was uniform. We 
then obtained by interpolation the values of the zero readings at the moments of 
making the deflection observations. The differences between these and the deflection 
readings gave then the true deflections right and left, q x and q. 2 respectively, the whole 
correction being a very small fraction of the measured deflection. 
The second series of throws were then treated in the same manner as the first, and 
a second pair of values of p x and p 2 obtained. These generally differed somewhat from 
the first, for the electromotive force of the battery—a combination of Daniell’s cells 
