191 1.] AGGREGATES OF ELECTRONS. 361 



view is taken the estimated frequency is not changed enough to 

 affect the general argument. 



If the Roentgen rays are so much Hke white hght you will at 

 once ask why they are not deviated by a prism. The answer is very 

 simple. It follows from the principles of the electron theory that 

 the refractive index /a of a substance for electromagnetic vibrations 



of frequency p is given by ^- ^ i -|- 2 — t — f 2\ "^vhere e is the 



charge on an electron, in its mass, />., its natural frequency and vs 

 the number of electrons of type .s^ in unit volume of the material. In 

 general this formula will not be exact on account of the interactions 

 of the electrons on one another but it will give results of the right 

 order of magnitude. Xow c and ;;; have the same value for all 

 electrons and in the part of the spectrum near the visible vgC-frnpa^ 

 is of the order unity for such frequencies /',, as fall within that 

 region. Xow r., will always be of the order of the number of mole- 

 cules per cubic centimeter whichever of the .y classes of electrons we 

 consider, so that we may draw the following conclusions : ( i ) On 

 account of the very great absolute value of p. ft- will be equal to 

 unity except for a very narrow range in the immediate neighbor- 

 hood of p = pg- (2) Only such substances will be capable of re- 

 fracting the Roentgen rays as have natural frequencies ps which lie 

 within the range of values of p embraced by the Roentgen rays ex- 

 perimented on. In any event it is clear that with a mixed group 

 of rays such as is emitted by an ordinary X-ray bulb, practically the 

 whole of them will pass through a prism without deviation. 



Barkla's experiments on secondary rays show that the Roentgen 

 rays exhibit phenomena very much akin to fluorescence in optics. 

 One interpretation of Barkla's results would be that there are in 

 material atoms natural frequencies comparable with the frecjuencies 

 in the Roentgen rays. In that case, although almost the whole of a 

 beam of Roentgen rays would be undeviated by a prism there 

 should be a small amount which would be deviated. At present I 

 am making experiments to detect this effect. 



Of course if one adopts the corpuscular view of the Roentgen 

 rays recently developed by Bragg, eft'ects of the kind described are 



