Ions of Alkali Salt Vapours in Flames. 713 



Thus if X' and n refer to the upper part of the flame 

 containing the salt, we have 



i = X V</h + h 2 ) = Xnefa -f k 2 ), 



which is independent of a, so that it does not appear at first 

 sight why the salt should increase the current with any 

 potential difference. 



If, however, X is big enough, the metal ions will move 

 down against the upward stream of gas and will be deposited 

 on the negative electrode. The alkali metal will consequently 

 accumulate at the lower electrode, and since it is strongly 

 ionized it will dimimsh the resistance there and so increase 

 the current. In the case of sodium salts this accumulation 

 can be easily observed by the appearance of sodium light 

 near the lower electrode*. 



It appears, therefore, that the increase in the current is 

 not as was originally supposed, due merely to the current 

 carried by positive ; ons coming down, for before the salt is 

 put in there are air jady present far more than enough ions 

 to carry the current allowed by the resistance at the negative 

 electrode. 



Suppose that an alkali metal atom in the flame is ionized 

 for a fraction f of the time. Then its velocity due to an 

 electric field will be fk{K. instead of £ 2 X. If, then, X 

 denotes the least value of X for which the metal accumulates 

 at the lower electrode, we have 



or „ u 



X 



fk x = 



for duriug the fraction (1—/) of the time the atom will be 

 carried upwards with velocity u. 



The quantity which was determined experimentally was 

 therefore not k x as was supposed, but fk x . Now the con- 

 ductivity imparted to a flame by equal numbers of molecules 

 of different alkali metal salts increases rapidly with the 

 atomic weight of the metal. This shows that a ca3sium atom 

 is ionized for a much larger fraction of the time than a 

 lithium atom. Hence, since both give the same value for fk u 

 it follows that k Y for lithium must be really much greater 

 than k 1 for caesium. 



In hot air at about 1000° C. the /'s will be much smaller 



* II. A. Wilson, Proe. R. I. 1900. 

 Phil. May. S. G. Vol. 21. No. 126. June 1911. 3 A 



