Reflexion of X Rays by Crystals, 887 



when they should, but other portions which lie off the band 

 are reflecting when they should not. The errors average 

 out. But there is an especial risk in using a wide pencil. 

 When the crystal is set at the fine glancing angles of first 

 order reflexion, five or six degrees it may be, there is danger 

 that some part of the pencil may not fall on the crystal face,, 

 unless the latter is very wide. It appears probable that in 

 earlier measurements the magnitudes of several first order 

 reflexions were underestimated on this account. 



When, however, the newer method is employed, the effects 

 of crystal irregularity largely disappear even when fine- 

 pencils are used. Each piece of the face on which the 

 pencil of X rays falls gets its chance in turn, and when all 

 is added np, it is as though the differently set pieces had 

 all acted together. This is a very important consideration. 



Although the crystal irregularities prevent all parts of it 

 from acting together, the reflected rays from all parts pass- 

 through the slit of the ionization chamber even when it is 

 quite narrow. This curious focussing effect has already been* 

 explained in the case of a perfect crystal *. It is dependent 

 on the equality of the distances from the crystal to the source 

 and to the slit of the ionization chamber respectively. It 

 may be shown that it holds good to a sufficient extent even 

 when the crystal is not perfect. This is a very convenient 

 circumstance, because the slit of the ionization chamber may 

 therefore be set in position to take in the reflected rays of a 

 given wave-length for all positions of the crystal, and at the 

 same time its width may be limited so that there is no chance 

 of other wave-lengths entering the chamber. For instance, 

 there are two strong rhodium lines, and it is easy to use the 

 one and not the other. The stronger line is really a doublet 

 (see below), and in the higher orders it is possible to use one 

 constituent of the doublet and not the other. 



One difficulty, apparently of minor importance, should be 

 mentioned : it is easiest to explain it by a numerical illustra- 

 tion. Suppose the glancing angle is 6°, and the crystal has 

 to be turned through the range from 5° to 7° in order to give 

 all parts of the crystal on which the rays impinge the chance 

 of reflecting. Then it is clear that this area is larger when 

 the angle is small (5°) than when it has the large value (7°). 

 Parts of the crystal are acting in the one case which are not 

 acting in the other. If the crystal were perfect, it must be 

 noticed, this would be quite proper ; but an irregular crystal 

 would only give an approximately true value through aver- 

 aging. It appears probable, however, that this effect is not 

 * Proc. Roy. Soc. lxxxviii. p. 433. 

 3N2 



