136 SCIENCE PROGRESS. 



This is, in fact, very nearly the case, for v 1 = 23*54, 

 v 2 = 14*65 and z\ + v 2 = 38*19, while the volume of the salt 



is 37-85- 



Evidently there are many ways of arranging the 

 atoms to accord with any required symmetry, and these 

 will all give different values of d 1 and d 2 ; the only 

 criterion which can be applied in order to discover which 

 is to be adopted in any instance is the proposed equality 

 between the sum of the partial volumes and that of the 

 salt itself. 



The insecure nature of the evidence is so conspicuous 

 that it is scarcely necessary to point out how at every step 

 in the reasoning a fresh assumption is introduced. We do 

 not know in the least that three molecules are necessary 

 for rhombohedral symmetry, or that the nitrogen atom is 

 attached to five oxygen atoms, or that it is allowable to 

 imagine the optical action of the nitrogen divisible into 

 fifths, or that there is any such thing as vertical or 

 horizontal linking, and least of all do we know the signi- 

 fication of the quantities d 1 and d 2 which are made to play 

 the part of densities measured along a certain direction. 

 Without other confirmatory evidence it is impossible to put 

 much faith in the criterion by which the atomic arrange- 

 ment is tested, or to say that the hypothesis possesses more 

 than a certain plausibility. 



But all this does not diminish the interest of these 

 suggestive speculations or the extreme ingenuity with 

 which they are brought forward by Sollas. 



An additional piece of evidence in favour of the distri- 

 bution which he suggests for sodium nitrate is the fact that 

 in potassium nitrate the least index of refraction is almost 

 identical with the lesser index of the sodium salt ; hence it 

 might be expected that this index is related to the radicle 

 which is the same for both salts, namely, the nitric oxide ; 

 and such a view accords with the atomic distribution 

 suggested above. 



Another case in which somewhat stronger confirmatory 

 evidence is adduced is that of the hydrated potassium 

 copper chloride Cu Cl 2 , 2K CI, 2H 2 O, which crystallises 



