223 



Almost any condenser when suddenly discharged if allowed to stand a 

 few minutes will experience a rise in potential at its terminals. This rise is 

 due to what is known as the residual charge. This phenomenon is explained 

 as follows : When a condenser is charged its dielectric is strained and being 

 non-homogeneous the strains are unequal. (By strain is meant the actual 

 compression of the plates.) When discharged these strains are i-elieved but 

 they do not decrease at the same rate, so that some parts of the dielectric- 

 become strained in the opposite sense and balance those parts which are 

 slower in acting. The condenseer is then apparently discharged, but after 

 standing a while these strains tend to diminish and usually there is a re- 

 sultant strain set up. This resultant strain is due to the fact that while 

 the forces were originally balanced at the end of the first discharge, yet 

 the distances are unequal and in nonhomogeneous materials stress is sel- 

 dom proportional to strain. 



The condenser may now be discharged again and after a time may 

 show still another rise of potential. In the apparatus herein described 

 this effect is entirely negligible, for the reason that the condenser i» never 

 charged or discharged suddenly, some few seconds being required to com- 

 plete the action. 



In all condensers there is also some absorption, but with good con- 

 densers used at the voltages proposed for this apparatus this effect is also 

 quite negligible, and we may with certainty say that for a given voltage 

 change at any part of the potential range equal quantities of electricity 

 pass through the instrument. 



With an instrument giving a uniform scale therefore we have an ap- 

 paratus which will show equal increments of readings for equal rates of 

 change of velocity, i. e. a direct reading accelerometer. 



Fig. 12 shows the apparatus as assembled for use in railway work. 

 The double commutator magneto is here shown belted directly to the car 

 axle. It is obvious that the readings of the instrument are unaffected by 

 grades or side tiltings of the car. 



The apparatus may be made self-recording by employing a recording 

 instrument instead of an indicating one, as shown in Fig. 13. These re- 

 corders may be obtained in the markel and arc very sensitive and reli- 

 able. The record is made by placing n pen on the end of the voltmeter 

 pointer, the whole being pulled down upon a sheet of paper moving at a 

 uniform rate of motion by means of a small magnet whose circuit is 



