388 Dr. A. LI. Hughes on the 



must set in sharply somewhere between X2002 and X2200. 

 There are no wave-lengths in the mercury-arc spectrum 

 within this region to enable a closer approximation to be made. 

 A linear relation between the maximum energy of emission 

 and the frequency was found to represent the results for 

 metals *. The law was expressed in the form 



V = kn-Y , 



where V is the maximum velocity measured in volts (i.e. the 

 potential difference required to bring the electron to rest), 

 n the frequency, and k and V are constants. The value of 

 k — the gradient of the line connecting energy with frequency 

 — does not vary much for metals, its mean value being about 

 3'6 x 10 ~ 15 . Putting this into the equation and the *87 volt 

 corresponding to the wave-length X 1849, we get V = 5'0 volts. 

 With these constants, k and V , we can calculate at what 

 wave-length the emission velocity is zero, in other words, the 

 wave-length at which the effect starts. The wave-length is 

 X2170, which is within the experimental limits X2002 and 

 X2200. Hence we can say that the relation between the 

 energy and the frequency is of the same order for anthracene 

 as for metals. 



The leaks corresponding to the wave-lengths XI 849 and 

 X2002 were much too small to enable me to determine the 

 nniximum emission velocities for the isolated wave-lengths, 

 as this is a much more difficult experiment than determining 

 whether a leak exists or not. 



Oxides of Barium and Copper. 



It may be interesting to place on record two experiments 

 on oxides. (These surfaces were not prepared by distillation 

 in vaouo.) 



A disk of copper was made red-hot in air so as to cover 

 the surface with a layer of oxide. It was then placed on the 

 nickel disk U of fig. 1, and exposed to the light in the same 

 way as the other compounds. Leaks of 4000 and 2600 were 

 obtained from two different surfaces. These were far larger 

 than the leaks from the salts, though they were of the order 

 of 100 times less than the leak obtained from distilled metals 

 under similar conditions. 



In the second experiment, the nickel disk was covered with 

 a thin layer of sealing-wax, and then strongly heated by the 

 blowpipe until a white deposit was left. This is the process 

 frequently adopted when making a " lime" cathode, though 

 the deposit is, however, mainly barium oxide. The leak was 

 never more than 4 units. The surface behaved like an 

 * Hughes, Phil. Trans. (A), vol. ccxii. p. 205 (1912). 



