370 SCIENCE PROGRESS 



of the beam. This latter is obtained by two slits, 5 cm. or 

 more apart. The powder is contained in a thin tube of a light 

 amorphous substance, such as glass or collodion, which should 

 be rotated if the grains are greater than about o-oi cm. diameter. 

 The trace of the cones is obtained on a film bent into the form 

 of a semicircle or quadrant with the powder at the centre. This 

 only includes the lines whose glancing angle 6 is less than 45°. 

 For special purposes and with softer rays such as the copper 

 series the whole circle may be necessary or a semicircle with 

 the primary beam as diameter. A " Duplitised " film with 

 intensifying screens saves time in exposure but is less exact 

 in definition and intensity. The diameter necessary depends 

 on the wave-length used, the width of the beam and of the 

 powder, and must be sufficient to obtain resolution of the lines. 



The width and indefiniteness of the lines obtained is a 

 considerable disadvantage to the accuracy of the resulting 

 determinations and an important and promising modification 

 has recently been described by H. Bohlin {Ann. der Physik, 

 5, 61, 1920). He uses a cyHndrical camera through a sht in 

 which a diverging beam irradiates the powder pressed into an 

 arc of the cylinder. All rays reflected at the same angle are 

 subtended by the same arc, and it can easily be shown that the 

 traces of the cones from all parts of the surface of the powder 

 due to radiation from the nearer edge of the slit are enveloped 

 by the image of that edge. All other reflections due to size of 

 slit and penetration of surface are included within this Hmit, 

 which appears as a sharp boundary. It seems rather a dis- 

 advantage that the proportional error is greatest in the case 

 of the most important lines of small glancing angle and that 

 the intensities are still less amenable to, interpretation than in 

 the original method. 



On development of the film, the distances of the lines from 

 the central trace are measured and 6 obtained for each reflection. 

 The order of the reflections can be seen, and a list made of the 

 spacings of the reflecting planes. The process is now one of 

 trial and error. In the case of metals and substances whose 

 crystal symmetry is simple and known, this is fairly easy. The 

 spacings of the planes, and the number of co-operating planes 

 for each spacing capable of giving a reflection, are calculated and 

 tabulated. This can be done once and for all for any lattice 

 system, altering only the scale for different materials. The fist 

 already obtained may then be compared with those for lattices 

 of the same symmetry. 



The possibilities and complications become overwhelming 

 with unsymmetrical compounds containing several types of 

 atoms, and none such have yet been analysed by this or any 

 other method. 



