( 440 ) 



m - — = e A 5 — re. -r 



de ^ dt 



Here /// is tlie mass, e the charjie ofl lie electron, 5 its displacement 

 parallel to the axis of A' from a ])ositioii of eipiilibi'itim, A' the com- 

 ponent parallel to this axis of the external electric force iTcting on 

 the electron; r and v>- are positive constants. 



In working ont the theory it a})pears that Yoigt's constant (//, is 

 identical with the expression y't>/4:T of Drude. Now the xalue of v> 

 was calculated by Drude himself (p. 490) from the vil)ratory j)eriod 

 of the sodiumlines; he iinds the value of this constant to be 7,G . lO^-"*; 

 from this value and that of V^oigt's constant a/, which I mentioned 

 just now, we lind : 



r = 1G50; 

 here we must bear in mind that this value applies to the ])articular 

 sodiumflame to which my measurements i-elate; r must, as well as 

 a/t, depend on the density of the sodinmvapour in the tlame. 



Drude tries in his book to tix limits, between which the value of 

 /' must lie. He tinds a lower limit by deducing from theory the 

 proportion between the quantity of light, which the absorbing sodium- 

 flame itself begins to emit under the influence of incident radiation, 

 and the quantity of incident light ^vhich is absorbed. This proportion 

 he finds to be 0,126//'. From the fact that reversal of a sodiumline 

 is i)ossible, he concludes that this })roiK)rtion must be considerably 

 smaller than 1, and he therefore fixes the lower limit for the value 

 of /■ by assuming: 



r>10. 



A higher limit is found by Drude from the consideration of the 

 phenomena of interference. He deduces theoretically the value of the 

 coefficient of danqjijig y of the free vibrations of the electron and 

 finds foi' this: 



7 = 0,(3 . r . 10-" . 



Now this coefficient nnist be small, as with gi-eat j)hase-difrerences 

 interference-phenomena can still be observed. With sodiundight inter- 

 ference-phenomena have been observed with a phase-difierejice cor- 

 responding with 200000 wavelengths; therefore 200000 y must still be 

 smaller than 1, therefore in this case: 



r < 100. 



It is evident that this result is not at all incompalibh,' with the 

 value of /' Avhi<-li I calculated above. In order to observe interference- 

 phenomena \Nilli such urcat phase-differences it has been necessary 

 to use a source of light showing very narrow sodiumlines; with 



