Kennedy — Large Ions and Condensation- Nuchi from Flames. 71 

 «£, then N = xiic , and consequently 



-dt = - '^'"^- 



where -^ means the collision frequency between ions of diiferent sign, every 



such collision involving the loss of charge of the two colliding nuclei, and 



resulting in the formation of one uncharged nucleus, with the subsequent 



history of which we are not concerned, since it is only the charged nuclei 



that are tinder consideration at present. If there are n nuclei per c.c, it has 



been found that 



dn 



that is, the collision frequency is -yw, each collision invohdrg the loss of one 

 nucleus. With equal numbers of positively and negatively charged nuclei, 

 or, 2nc altogether, the collision frequency will be 47?;^". If the collision 

 frequency be independent of the charge, as seems true from the experiments, 

 the number of collisions per second taking place between the positively 

 and negatively charged nuclei will be half the entire number, or 2-y7i/. 

 Therefore 



dn. 



dt 



jnc, 



and consequently 



^X ^ 2y. 



Now 



|3 = 6-3 X 10-1°, and y = 14 x lO-'", 

 therefore 



X = 4"5, 



and this value agrees well with the values obtained under similar circum- 

 stances from direct measurements of N, n, and Uo, the calculations being made 

 from the decay curves of these quantities. 



Tfie Mobility of the Large Ion. 



The resvdts given in this paper show that the charge on the large ion may 

 have widely different values under different circumstances of production. 

 Even in any one mass of flame-gas it is not likely that the values of the 

 charges borne by the different ions will be the same, though they will 

 probably vary but Little round a mean value. When the gas has been 

 deionized by an electric field, and the nuclei charged again by ionizing the 



