2QO 



NATURE 



[October 22, 19 14 



is ftlled with some heavy gas, so that the ionisa- 

 tion current may be as large as possible. Sulphur 

 dioxide is convenient in many cases, but methyl 

 bromide is much better for rays which can excite 

 the bromine X-rays. Such rays are, for example, 

 given olT by antikathodes of silver, rhodium, or 

 palladium, the latter two of which have been much 

 used iij the determination of crystal structure, 

 because they give off intense homogeneous rays 

 and also stand up well against the bombardment 

 by the kajthode stream. The chamber I is insu- 

 lated and maintained at a high potential, which 

 drives any ionisation on to an internal electrode. 

 The latter is connected by a fine wire which passes 

 down inside the metal shield \\'\V to the gold 

 leaf electroscope (Wilson pattern) within the 

 shield E.^ • The connection with the electroscope 

 terminal is made at a point lying on the axis of 



Fig. 2. — Photograph of X-raj- spectrometer. 



the spectroscope, so that the ionisation chamber 

 I, the casing W, and the connecting wire may all 

 revolve together about the spectrometer axis 

 without straining the connection. The shielding 

 is made solid and strong : it is necessary that the 

 electrostatic screening should be perfect, and the 

 electroscope must be protected from draughts 

 which may cause changes in temperature. 



The gold leaf is illuminated by reflection from a 

 mirror L, and viewed through a microscope M. 

 A strong X-ray reflection will cause the leaf to 

 move twenty or thirty scale divisions in a second. 



The angular positions of the crystal and the 

 ionisation chamber are read from verniers at V 

 and V^ Ob.servations can easily be made to half 

 a minute of arc, and much finer work could cer- 

 tainly be done, if it were required. The actual 

 angle of reflection can be measured with an 

 accuracy higher than can be reached in our know- 

 NO. 2347, VOL. 94] 



ledge of certain other data used in some of the 

 calculations ; for exarnple, the actual weights of 

 the atoms. 



Crystals are often very imperfect in construc- 

 tion, consisting rather of a conglomerate of 

 smaller crystals in more or less imperfect align- 

 ment. It is interesting to observe that the spec- 

 trometer may be used in ways which almost com- 

 pletely overcome the evil effects of the imperfec- 

 tions. In the case of a very perfect crystal like 

 the diamond, the slit at A is not used, but B is 

 set very fine ; D is wide open. The crystal is 

 slowly turned by a tangential screw attached to 

 the revolving table, and there is no reflection at 

 all outside very narrow limits. The angle of 

 maximum intensity of reflection can easily be 

 determined to a few seconds of arc. But a crystal 

 .of rocksalt cannot be treated in this way. It is 

 \ • , best to set A and D fairly fine 



and not to use B at all. On 

 account of certain most for- 

 tunate geometrical considera- 

 tions, a homogeneous pencil of 

 some divergence issuing from A 

 and reflected at various points 

 on the crystal face is brought 

 to a line focus at D, provided 

 that A and D are at equal dis- 

 tances from the crystal (Proc. 

 Roy. Soc, Ixxxviii., p. 433)- 

 A perfect crystal would re- 

 flect such a pencil only along 

 a certain vertical line on the 

 crystal face; but a poor crystal, 

 like rocksalt, at a number of 

 separate points on the face. Even 

 crystals which are scarcely re- 

 cognisable as such may be 

 treated by this method. 



The higher orders of spectra, 

 that is to say, reflections at 

 angles for which n has a 

 large value, three, four, or 

 five, naturally give more ac- 

 curate values than lower orders, 

 though the intensity dimin- 

 ishes rapidly as n increases. The 

 "resolving power" increases even faster than n, 

 since dOidX is easily seen to be equal to tan^/X, 

 which becomes very large as 6 approaches the 

 value ■7r/2. For example, a certain pair of lines 

 emitted by a platinum antikathode are separated 

 by thirty minutes of arc in the first order spectrum 

 reflected by the cleavage face of the diamond, but 

 in the third order spectrum they are two and a 

 half degrees apart. 



With a little practice it is quite easy to pick up 

 the reflected X-ray. While the search is being 

 made the slits are opened wide ; as soon as the 

 reflection is ob.served the slits are made narrow, 

 and accurate measurement is then possible. It 

 is a comparatively simple matter to find the 

 angles of reflection of rays of given quality in 

 the various faces or sets of planes of a crystal. 

 The greater difficulty arises in the geometrical 

 interpretation of the results. 



