Manchester Hfevioirs, Vol. Ivii. f'1913), No. Vi,. 3 



It will be seen froin my gf.-ncral Table that ihc atoirii<: 

 weights of nitrogen, silicon, and iron, besides bcinc^ wliolc 

 numbers, are exact multi[jles of II7; and in all the 

 formul.e proposed for the constitution of silica the atomic 

 weight of silicon is a multiple of 7. These formul.x- arc 

 given below, with the rjlrl and new atomic weights, the 

 proportion of silicon to oxygen being in the ratio of 7 : <S 

 in all the formula;. 



r. SiO =Si 7 :0 8 : : 7 : « 



2. Si02 = Sii4:Oi6::7 :8 



3. Si02 = Si28 :032 : :7 :8 



4. Si 0:, = Si2i :024: :7 :8 



5. SiO,, = .Si42 :048: :7 :8 



6. Si20f, = Si35 : O40 : : 7 : 8 



7. Si,Or, = .Si70:C)8o: :7 :8 



I have shown that the rndiiial number lA the typical 

 molecule at the hearl of the several series in the general 

 Table determines the quantivalencc of each series of 

 elements under it. When my first paper on atomic weights 

 was published the only member of the series Wyn known 

 to be heptavalent was manganese, but I therein stated 

 that the relation of this element of the iron group indi- 

 cated a much liigher qiiantivalence for the other members 

 of this series than had iiitherto been accorded to them. 

 MM. Hautefeuille and Chappuis have since formed per- 

 nitric acid, which indicates a higher quantivalence for 

 nitrogen than had previously been obtained for this 

 element ;* and more recently MM. Debray and Joly have 

 shown that ruthenium fkuj is heptavalent by ilic forma- 

 tion of the hcptarutheneatcs of potassium and sodium, 

 which have many points of resemblance to the heptaman- 



ganates.f 



* Couiptes Rendu-,, tome 94, \>\k i j i r, 1306. 

 t- Caiiiptea A'eiidiis, tonic 106, \>\>. 1494, 1888. 



