46 



PROCEEDINGS OF THE AMERICAN ACADEMY 



Co, 



CI 



NIL- 



NII.,'- 



Nir,'- 



Nir!- 

 a 



-NIL— Cl 



-NIL— NH..— CI T ^ J 

 -NH -Nil -CI ^^^^ ^^-' < 

 -NH— CI 



NIL— CI 



NH'— NIL,— CI 



N H,— NH — N H,— CI 



NH,— NH,— NH.!— CI 



NH3— NH'— CI 



NR3— CI 



As I have adopted this view throughout, further ilhistrations will be 

 unnecessary. The advantages of the theory are, I thiuk, first that it 

 regards cobalt as tetratomic, and reduces the whole series virtually to 

 the type of CogO^, or CogClg, and secondly that it enables us to explain 

 the different cases of isomerism without arbitrary assumptions as to 

 the existence of allotropic forms of cobalt. Thus it is easy to see that 

 there may well be a difference between compounds having respectively 

 the symmetrical structural formulas, — 



I. Co., ^ 



NILj— CI 

 NIL.— NH3— CI 

 NH.;— NH.,— CI 



NH.,— NHg 

 NIL— NII3 



NII3— CI 



-CI 

 -CI 



II. Co, ^ 



CI 



NH.,— NH.,— CI 



N H.!— N Hi— N IT.,— CI 



N I L— N H3— N H3— CI 



NHl— NIL— CI 



CI 



III. Co„ i 



CI 



NIL, 



NIL 



-CI 



-N Hg—NHg— NH3— CI 



NH3— NIL,— NH3-^ 



NH3— CI ' 



CI 



-NH,— CI 



IV. Co., 



r ci 



Cl 

 NIL 



NH3— NIL 



Cl 



Cl 



.NH3— NH3— NH3— NIL,— Cl 



-NH3-NH3- 



-NIL— Cl 



and that supposing that the six units of affniity of Co^ are quitlitatively 

 exactly equal, the complex, Co^,(NIL)j|3Cl(;, admits of being arranged 

 in a great variety of ways. Now since the salts of roseocobalt and pur- 

 pureocobalt resemble each other as regards the types to which they 

 belong and as either of these types may be formulated in a great 

 variety of ways, it becomes very difficult to determine which structuial 

 formula to adopt for the com[)Ounds of either series. 



Thus the salts of roseocobalt and of purpureocobalt, usually at least, 

 belong either to the types ReX^ and PcX,., to'RcX^Y^ and PcX^Y^, 



