FL UCTUA TIONS IN MICROPHONES A ND OTHER RESIST A NCES 219 



resistance modulation due to the cyclical compression is independent 

 of the applied voltage. It is evident from Eq. (1) that the fluctuating 

 resistance responsible for the noise cannot be equivalent to the resist- 

 ance modulation introduced by a cyclical compression where the 

 contacting granules move relatively as a whole, for the fluctuating 

 resistance responsible for noise is somewhat voltage sensitive as indi- 

 cated by the departure of a from the value 2. This means either that 

 the conductance responsible for the noise is specifically non-ohmic or 

 that the extent of the conduction wherein the noise mechanism lies is 

 diminished as the applied d.-c. voltage is increased. Non-ohmic 

 conductance is usually such that conductance increases with applied 

 voltage, thereby demanding a value of a in Eq. (1) which is greater 

 than 2; accordingly we are inclined to believe that applied voltage acts 

 to diminish the area over which the noise mechanism operates. 



If the noise mechanism were intimately associated with the total 

 conductance of a contact one would expect the noise to be proportional 

 to some simple integral power of the current in a contact, but this is 

 denied by the observed value of in Eq. (2). 



These facts and deductions lead us to the hypothesis that there 

 exist two mechanisms of conduction between particles in contact, a 

 primary conduction which accounts for the major portion of the 

 current, and a secondary conduction wherein a relatively small por- 

 tion of the total current is transferred and in which the noise mechanism 

 is found. Goucher ^ has given evidence that the primary conduction 

 between contacting carbon particles is of the same nature as that in 

 solid carbon, and since we have been unable to measure any noise in 

 solid carbon we assume that the secondary conduction does not take 

 place through the same region of the contact as the primary conduction. 

 Recent investigation of the elastic nature of carbon contacts ^ has 

 led to the conclusion that the surface of each particle can be considered 

 as covered with a layer of hemispherical hills of heights distributed 

 according to the function N^ = Const. x'\ where N^ is defined as the 

 quantity which when multiplied by ^.v gives the number of hills coming 

 into coincidence with a plane as it moves from the position x to x + dx, 

 and 71 is a constant whose experimentally determined value is about 

 0.6. The establishment of a contact consists in bringing into coinci- 

 dence a number of these hills and enlarging the coincidence areas to the 

 extent demanded by the displacement of the contacting elements after 

 their initial coincidences. Let us accept this picture of a contact and 

 inquire as to how it applies in the explanation of our empirical noise 

 equation. 



We assume that through each area of coincidence the primary con- 



