Inductive Capacity of Dielectrics, 531 



must be multiplied by its distance y from the centre, and sum- 

 mated between points situated approximately at the ends of 

 the needle and at the inside edge of each condenser-plate. 



< 



V*(K-K l)a 



where 21 is the length of the needle, and %' the diameter of 

 the central opening for the bi filar suspension. 



11171. 



This is to be equated with Wg -j~ sin <£, the torsional moment 



of the bifilar suspension ; W being the weight of the needle, 

 g the acceleration of gravity, m and n the upper and lower 

 distances apart of the threads, h their length, and </> the angle 

 turned through. 



Thus the sine, or, if the deflexion is small, the angles turned 

 through should be proportional to the square of the electro- 

 motive force applied to the plates. This was in complete 

 agreement with our experiments on sulphur. Some experi- 

 ments, however, made with a light ebonite needle gave angles 

 of deflexion lying between the square law and a directly pro- 

 portional law. This discrepance was probably due to the 

 unreliability of the bifilar method of suspension. 



To calculate K, we have 



The mean of several sets of experiments on sulphur gave 

 the value K = 2*56 ; but as our object was more to demon- 

 strate the practicability of the method than to value K, we 

 did not make arrangements for very accurately determining 

 the quantities involved, and consequently we can put no great 

 reliance on this number. Indeed, without a careful physical 

 and chemical examination of the specimen of the substance 

 under examination, a great degree of refinement is of little 

 use, a fact too often lost sight of. 



A considerable time-lag was always noticed on reversing 

 the polarity of the condenser-plates. That is to say, the spot of 

 light would go back towards zero a few divisions on reversing 

 and then slowly creep up to the former value. It ultimately 

 always came up to approximately the same value as before 

 reversing, thus exhibiting little or no dielectric hysteresis. 



By applying a rapidly alternating source of electromotive 

 force to the condenser-plates an approximation towards the 

 " instantaneous value " of K may thus be obtained. In fact 

 the method is particularly suited to this purpose. Some 



