334 Messrs. Guthrie and Boys on 



results differed by ouly 1/3 p. c. This result seemed very 

 satisfactory, especially as possible changes in the strength of 

 the magnet or in the torsion of the wire were not considered ; 

 and therefore we determined that the experiments should be 

 continued with the old wire. So it was well rubbed with 

 sperm oil to stop the rust from increasing in quantity and 

 destroying the wire. 



The method of carrying out the experiments must be de- 

 scribed next. It has already been shown that the magnets 

 when at rest have a directive action on the cell, but when in 

 motion their action on the cell itself is nothing ; yet if a con- 

 ducting liquid is in it the cell is turned round. Also the 

 zero of the scale can only be determined after the liquid is 

 put in ; and therefore there is no possibility of observing the 

 zero directly. But as the torsional effect is proportional to the 

 speed, by running the magnets first at a low and then at a 

 high speed and taking the difference in the speeds (S) and 

 also the difference in the readings (T), and dividing S by T, 

 a number is obtained which is a measure of the resistance of 

 the liquid in the cell. To diminish as much as possible all 

 chances of error, every liquid was examined at four speeds — 

 two with the magnets turning in one direction ( + ), and two in 

 the other direction ( — ) ; and the results were taken in pairs, 

 in the order of observation, to avoid errors due to such slow 

 changes in the zero as were still liable to take place. If the 

 strength of the magnets and the torsion of the wire were both 

 constant, nothing more would be required ; but as both are 

 liable to change, some standard of comparison by which the 

 value of an observed deflection can be measured is necessary. 

 The most obvious plan is occasionally to examine some given 

 liquid of good conductivity, e. g. 30 p. c. sulphuric acid, and 

 consider variations in the result due to changes in the ma- 

 chine, and correct the measures obtained for other liquids 

 accordingly. But this plan would involve a great waste of 

 time, as from two to three hours are required for the exami- 

 nation of a liquid ; and it would not be trustworthy, for it 

 cannot be supposed that the torsion of a wire is the same when 

 stretched to 50 p. c. and to 80 p. c. of its breaking strain ; and 

 therefore the standard of comparison must be applied at every 

 experiment when the liquid under observation is in the cell. 

 Accordingly a disk of thin sheet brass, D, three inches in 

 diameter, was used as a comparison plate. It is hung to the 

 glass hook below the beam B, immediately after the obser- 

 vation at the fourth speed has been made, without stopping 

 the engine or touching any other part of the apparatus, and 

 the increase in torsion noted. This increase in the torsion 

 when corrected for speed should be constant ; and the slight 



