STRUCTURE OF CRYSTALS — WYCKOFF. 215 



procedure is, as we have already pointed out, that those original 

 structures, upon the correctness of which the value for the wave 

 length of X rays has been based, have never themselves been shown 

 to be the only structures which are in agreement with the experi- 

 mental data. The strong indication of correctness given to this 

 value of the wave length by the fact that it fits in so satisfactorily 

 with the large amount of data that has since accumulated makes this 

 objection now a more or less formal one. 



The determinations thus far made of the structure of crystals 

 have been carried out by two distinct methods of procedure — or 

 perhaps it would be better to say from two different points of view. 

 Most of them have followed closely the early determinations in 

 searching around for some grouping of atoms which would account 

 for a certain limited amount of experimental data. The structures 

 thus obtained have been considered to represent the actual arrange- 

 ment of the atoms within the crystal. In many cases they are the 

 simplest arrangements which could be found and as such have a 

 certain, but very doubtful, degree of probability. Without excep- 

 tion, probably, they are not, however, the only ones which could be 

 fitted into the existing data. 



Only a few crystals have as yet been studied, starting from the 

 point of view suggested by the theory of space groups. Following 

 such a method of procedure we would most ideally make use of all 

 three ways of getting diffraction effects — the spectrometer, the 

 method of powders, and the transmission (Laue) photographs. The 

 first step would get a reflection spectrum from some important 

 crystal face. This measurement will decide the number of chemical 

 molecules (or usually the numbers, since more than one is possible 

 because of the uncertainty as to the order of the reflection) that can 

 be associated with the unit underlying the crystal. With this as a 

 basis and using the results of the theory of space groups, all of the 

 ways in which the atoms of the compound could possibly be ar- 

 ranged can be written down. It is then a comparatively simple mat- 

 ter to decide exactly what data are required to distinguish between 

 these various possibilities and to get them by whichever of the three 

 methods of obtaining diffraction effects will supply them most readily. 

 Even if it is impossible, with the present limited amount of infor- 

 mation about the mechanism of the reflection of X rays, to eliminate 

 all but one of these groupings, still we always have exact knowledge 

 as to the status of the determination. The element of guessing is thus 

 quite definitely eliminated. 



