142 



for (lie number of particles for which tlie vefocity § lies between g 

 and § -|- (l§, and the change of frequency between the corresponding 

 values to and w -\- day, 



1/ 

 1/ 



(16) 



dü} (17) 



This last expression immediately determines the distribution of 

 light in the emission line. The borders of the line may lie taken to 

 correspond to the values of w for which the exponent becomes — 1, 

 i. e. to 



I y^ ^ 



V 3 c 

 so that the width is determined by 



o 



If u is of the order of 5. JO' cm/sec and A„ of the order of 6000 A.U., 



o 



this AA will be about '/eo A.U. This is a very small width ; yet, 

 it far exceeds the value which, starting from the value of g^, 

 we found (§ 4) for the breadth of an absorption line, and which 

 would also belong to an emission line, if we had to reckon with 

 the radiation resistance only. The cause of the difference is that 



g,«mn,'- (18) 



c 



The conclusions drawn from (17) about the width of the lines are 

 ill good agreement with the results of several physicists ; they are 

 strikingly confii'med by the experiments of Buisson and Fabry') on the 

 emission of helium, krypton, and neon in Geissler tubes. These 

 observations show at the same time that in these rarefied gases there 

 are no resistances whose coeftlcient does not fulfil the condition 

 (18), and which, acting by themselves, would therefore give rise to 

 a width comparable with that arising from molecular motion, or 

 greater than it. If there had been resistances of this kind, the 

 observed widlh would have been found greater than is required by 

 Dopi'LKk's principle. 



1) H. BuissoN ct Ch. Fabry, La largeur des raies spectrales et la theorie cine 

 tique des gaz, Journal de Physique (5) 2 (1912), p. 442. 



