ARTICLES 57i 



Very naturally much of the recent work has been devoted 

 to the detailed improvement of the spectrometer with the 

 object of attaining more accurate measurements. A series 

 of excellent papers on the subject has appeared during the 

 last year or two in the American Physical Review. 



Lack of sharpness in the image, or, more generally, of " re- 

 solving power," is due mainly to the following causes : 



(1) The slits must be wide enough to pass sufficient X-ray 

 energy, or else the apparatus is too insensitive. 



(2) The diffraction effect which gives rise to the reflected 

 beam of X-rays is not a simple reflection occurring at 

 the crystal surface, but a volume effect in which the 

 whole of the crystal takes part, so far at least as the 

 X-rays penetrate. 



So far as has been observed no lack of sharpness is due to 

 an insufficiency in the number of waves in the X-ray train, an 

 effect which might have been supposed to be a possibility. 

 Also, crystals can easily be found whose accuracy of con- 

 struction is so good that no effects due to the lack of it can be 

 discovered. 



As regards the first of the difficulties stated above, it is 

 clear from fig. 1 that the width of the image must be at least 

 as great as the width of any slit which limits the width of the 

 beam of X-rays. Such a slit is usually set at a width of from 

 two- to three-tenths of a millimetre. 



As regards the second difficulty, the depth to which the 

 X-rays penetrate in the crystal adds further to the width of 

 the image ; and this effect can only be diminished by using 

 a very thin crystal. Taking these two points together, the 

 definition of the image can only be improved by making the 

 slit narrow and the crystal thin. Every step taken in this 

 direction makes the image weaker and renders it advisable to 

 employ as strong a source of X-rays as possible. 



The limits which are set in this way make it difficult to 

 separate lines which are close together ; the resolving power 

 of the instrument is small. In a case given by Dershem 

 {Physical Review, June 191 8), it was found that the resolving 

 power was considerably less than 200, that is to say, it was 

 not possible to separate two lines whose relative frequencies 



