Manchester Memoirs, Vol. xli. (1897), A r o. 15* 3 



the place of meeting of the lines by which the shadows 

 are formed, prolonged backwards into the tube, is the 

 place which is the seat of action of these rays ? I 

 think we have not. If a portion of the Crookes' tube 

 which is influenced by the cathode discharge be isolated 

 by, we will say, a lead screen containing a small hole, 

 you get a portion of the cathodic rays which come out 

 through that small hole, and you can trace what 

 becomes of them beyond. It is found that the influence 

 is decidedly stronger in a normal direction than in 

 oblique directions. Professor J. J. Thomson, of Cam- 

 bridge, who has worked a great deal experimentally at 

 this subject, mentioned that to me as a striking thing. 

 You might imagine that the fact that the shadows 

 appear to be cast approximately from a source within 

 the tube could be accounted for in this way. Supposing, 

 as Rontgen believed, that the seat of the rays is in the 

 place where the cathode discharge falls on the surface 

 of the glass, those which come in an oblique direction 

 have to pass through a greater thickness of glass than 

 those which come in a normal direction. Now, glass is 

 only partially transparent to the Rontgen rays ; therefore 

 the oblique rays would be more absorbed in passing 

 through the glass than the rays which come in a normal 

 direction. I mentioned that to Professor Thomson, but 

 he said he thought the difference between the intensity 

 of the rays which come out obliquely and those which 

 come out in a normal direction was much too great to 

 be accounted for in that way.*" I will take it as a 



* I have found by subsequent inquiry that the experiment referred 

 to was not made by Professor Thomson himself, but by Mr. C. M. 

 McClelland, in the Cavendish Laboratory, and that on being recently 

 repeated with the same tube the effect of the X rays was found to be 

 by no means so much concentrated towards the normal to the wall of 

 •the tube as in the former experiment. It seems likely that the difference 

 may have been due to use of the tube in the interval, which would 



