472 



Messrs. A. W. Porter and D. K. Morris. [Mar. 21, 



viscous effect as before ; the point U is reached during the throw of 

 needle: after which, a further viscous flow takes place, until (if the 

 cyclic state be set up) the point B is returned to once more. 



The contact B is moved back to A while the condenser is still 

 detached from it ; and this series of operations is then many times 

 repeated. 



Tf the dissipation of energy in the condenser is due to hysteresis, 

 then, as in the magnetic analogue, we know that the cyclic curve 

 will embrace an area, and the point Q will not coincide with the 

 point P ; and the converse of this is also true if time has been given 

 for viscous effects to subside. 



If, on the other hand, experiment shows no difference in the 

 charge when at Q from what it is at P, there can be no hysteresis, 

 unless of an amount so small as to elude this method of detecting it. 



The following figure (Fig. 3) shows the kind of curve that will be 

 traced if hysteresis be absent. 



FIG. 3. 



Potential difference. 



The experimental results are given in Table I. 



The point P was so chosen that the charge was then almost exactly 

 half of the maximum charge. The duration of a cycle of operations 

 was never less than five minutes. The galvanometer throws were 

 read by a telescope and scale, and are expressed in centimetres. As 

 both galvanometer mirror and telescope are of fine quality there is no 

 difficulty in detecting a tenth of a millimetre on the scale. 



It will be observed that the throws fall off gradually throughout 

 the experiment in consequence of gradual failure of the battery 

 power; but as this affects both series to the same extent it has no 

 influence on the ratios. Individual values of the ratio Throw at 

 Q/Throw at P differ amongst themselves by small amounts, being 

 belcw unity in 11 cases, above in 12, and equal to unity in one case. 



