OF ARTS AND SCIENCES. 35 



Bait, Co2(NH3)^(N02)8Ag2 ; it also gives, though somewhit sluggishly, 

 the characteristic reactions of salts of luteocobalt. The analA^ses and 

 reactions leave no doubt as to the true constitution of the salt. Its 

 relations to the other bodies metameric with it may be seen from the 

 expression, — 



^Co^CNH,),, \ \ Co,(NH3),(NO,) J3 = 4 Co,(NH3),XNO,),. 



It has the same molecular weight as the octamin salt : — 



lCo,(NO,),,nCo,(NH3),(NO,),^3. 



In the metameric series to which I have directed attention, at least 

 two other members are theoretically possible. Thus we should cer- 

 tainly expect the reactions and products indicated by the equa- 

 tions : — 



3.Co,(NH3),,(NO,),CI,+2.Co/NO,)„Na,3 = \ Co,(NH3),„^3 

 3.Co2(NH3)3(NO,),Cl,+Co,(NO,),oNa,, = ^€o,(NH3),(NO.,) J3 



The first or xanthocobalt salt would be empirically, 5.Co2(N.H.j)g 

 (NOo)g, while the second or croceocobalt salt would be 4.Co./NH3)g 

 (NO^)^. I have more than once been fully confident that I had ob- 

 tained both these salts ; but in the final revision of my work I did not 

 succeed in obtaiuing either for analysis, and their existence must 

 therefore, for the present, remain doubtful. The difficulty in prepar- 

 ing the salts of the Co2(NOo),2 series depends mainly upon the fact 

 that it is indispensable to avoid an excess of sodic nitrite in preparing 

 the solution of Co.,(N0.2)v2^^<j "^ that salt, as the sodic nitrite acts 

 readily on the new salts formed. The jiossible existence of the anhy- 

 drous xanthocobalt salt of the Co2(N02)j2 series is, however, shown by 

 the existence of the compound 



\Co,{NR,),,(m.;),UCo,(m,)j,+^on,, 



which I have described with the salts of purpureocobalt, but which, ns 

 already stated, belongs, as its reactions show, to the xanthocobalt series. 

 The metameric compounds, the existence of which may be consid- 

 ered as fully established, are as follows, denoting ammonia by A, and 

 NO2 by X, for brevity : : — 



