242 MESSRS. R. T. GLAZEBROOK AND J. M. DODDS ON THE 
oscillations of the galvanometer needle could easily be reduced to a few tenths of a 
millimetre of the scale. 
When this had been accomplished, the battery current in the coil A was reversed 
and the first throw of the galvanometer needle observed. Suppose this was a throw 
to the right. By reversing the current at the right instant the needle could be 
brought very nearly to rest again, and the small swing that remained was easily 
destroyed by the damper. The connexions were then adjusted so that on again 
reversing the battery current a throw to the left was observed. A second throw to 
the left was observed and then a second throw to the right. 
After this another reading was taken of the resting point when no current was 
passing. 
The connexions were then altered so that a fraction of the direct current could be 
passed through the galvanometer, and the position of rest of the needle when the 
current was passing was observed. The needle was easily brought sufficiently nearly 
to rest in its new position by making the primary contact for a third of the time of 
swing, then breaking it for a second third, and finally making it again and leaving it 
marie. 
The resting point was determined while the needle was swinging from the observa¬ 
tion of five consecutive turning points. The primary current was then reversed and a 
deflection in the other direction observed. After this a third reading of the resting 
point without any current was taken. 
The connexions were again altered to observe the throws and four more were taken, 
one to the left, two to the right, and one to the left. A fourth observation of the resting 
point completed this part of the observations. The secondary circuit was then put into 
communication with the Wheatstone's bridge, and the difference between It and It 
measured, giving a second series of values of x y, x and y', and finally the thermometers 
were all read again. 
When we had become accustomed to the work a complete set of observations, 
excluding the time of swing, took about 25 minutes. 
In the first series of measurements the time of swing was only observed twice each 
afternoon—at the beginning and end of the afternoon’s work. In the'second series it 
was taken generally at the beginning of the work and after every second or third set 
of observations, that is to say, at intervals of somewhat less than an hour and-a-half. 
Throughout the experiments one observer (II. T. G.) read the galvanometer deflec¬ 
tions, while the other (J. M. D. in the first series, E. B. S. in the second) made or broke 
the various connexions as required, and noted down the scale readings as they were 
read out by the observer at the telescope. 
To obtain from the direct results of the observations the quantities required for 
substitution in formula (5), the following method was adopted :— 
The means of the temperatures at the beginning and end of the observation were 
