REED — RELATION BETWEEN VALENCE & ATOMIC WEIGHT. 675 



6C/ equivalents to change selenides to selenites. 



6C/ " " tellurides to tellurites, 



2CI " " sulphites to sulphates, 



2C/ " " selenites to selenates, 



2CI " " tellurites to tellurates. 



4C/ " " sulphur to sulphites, 



4C/ " " selenium to selenites, 



4C/ " " tellurium to tellurites. 



6C/ " " sulphur to sulphates, 



6C/ " " selenium to selenates. 



6C/ " " tellurium to tellurates. 



These facts are too well known to need further authority, and 

 may be found in the equations of any treatise on chemistry. 



C/, Br and I are acknowledged to be monads in all chlorides, 

 bromides, and iodides ; pentads in all chlorates, bromates, and 

 iodates ; heptads in perchlorates and periodates. If the valence 

 of C/ is -{- I in chlorides, -(-5 i" chlorates, and -f- 7 in perchlo- 

 rates, the difference between its valence in chlorides and in chlo- 

 rates would be 4 ; chlorides and perchlorates, 6. 



But we find from the above list that these differences, as found 

 by experiment, are respectively 6 and 8 instead of 4 and 6. We 

 are forced to the conclusion, therefore, that the valence of C/ in 

 chlorides is not -\- i but — i, which makes the differences 6 and 8 

 as required. 



The same is seen to be true of all the elements enumerated 

 above. 



What further proof do we need of the existence of negative 

 valence? 



The validity of the hypotheses and arguments given above evi- 

 dently rests upon the truth of the general considerations which 

 have led to the adoption of our present atomic weights. The con- 

 clusion reached is, therefore, an expression of these considerations. 



Burlington, la., Oct. lO, 1885. 



