52 



SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 65 



tive elements of groups VI and V of the Periodic Scheme. Thus 

 in the short periods the effective numbers of magnetons will be 

 nearer to i 2 3 4. 3 2 i 



than to 1234567, although this almost certainly under- 

 estimates the numbers in groups V-VIL Similarly, in the long 

 periods we can substitute for the actual numbers, which are 

 123456789 10 (11) 



(1)2 3 4567, the effective 

 numbers : 1234567876 5 4321. The use of 



these numbers (in italics) would place C and Si at the minima of 

 the curve in the short periods, and Fe Co Ni and the Platinum 

 metals at the minima in the long periods — as they are in the curve 

 of actual atomic volumes. 



With regard to the shape of the curve, since the envelope, as we 

 have pictured it, is bound to be more compressible at small than at 

 great compressions, we could predict the flat minima and peaked 

 maxima that are observed, and would expect, from what has been 

 said, a curve of the general shape shown in the figure (the signifi- 



cance of the " dotted " loci is explained below). The chief defect in 

 this curve is that it places all the inert elements at the maxima. In 

 actual fact He is at the first maximum, and Ne probably at the second 

 (its density in the liquid state has apparently not been determined), 

 but the other maxima are occupied by the alkali metals K, Rb, Cs. 

 Now although it is difficult to explain why the atomic volumes of 

 these elements should be greater than those of A, Kr, Xe, it is easy 

 to see why they should not be very much less. When each atom in a 

 cluster has only one valence magneton, the chance that the valence 

 magnetons will cooperate effectively to compress the cluster is very 

 small, and much less, for example, than one-fourth of the chance 

 when each atom contains four — not only because of their small 



