48 



PROCEEDINGS OF THE AMERICAN ACADEMY 



and among the salts of purpureocobalt we find the formula, — 



Co,(NH3),„(N03)3CL+Co,(NH,),,(N03)«. 



Roseocobalt and purpureocobalt may therefore also form compounds 

 in which the six units of affinity are saturated by threes. Now since 

 tliese saks contain only ten atoms of ammonia; and since these can only 

 be distributed in pairs of different structure, as in formulas I., II.. 111., 

 and IV., it follows that we cannot fairly draw the inference that in the 

 dodecamin series the atoms of ammonia are arranged in six perfectly 

 equivalent pairs. 



Blorastrand gives to chloride of xanthocobalt the structural for- 

 mula, — 



f 0— NO 



a —a— CI 



a —a— a —CI 



a —a— a —CI 



a —a— CI 



O— NO 



Co., 



upon the ground that cobalt unites with O.NO in Fischer's salt 

 Co.,(N02)i2lVg with peculiar energy. We seem in this way to gain a 

 Ttov Gt(6 for this series ; and, if we admit the force of the argument, we 

 must write the formula of chloride of roseocobalt, — 



C02 < 



On the other hand it is a question whether the remarkable stability 

 of Co.,(NO.,),.,Kg can fiiirly he attributed to any special allinity nf 

 cobalt for NO.,. We have a complex whole, which we find remark- 

 ably stable ; but is not this stability the resultant — so to speak — of 

 the whole structure, just as the strength of an arch resides in the whole 

 arrangement of its elements, and not in any single one ? The otlur 

 c )l)alto-nitrites, as, for instance, Sadtler's salts, Co(NO^,)o-|-NaNO^,. and 

 Co(NO.,)2-|-2NaNO.„ are not remarkably stable, but ratlicr tlie re- 

 verse. The salts of xanthocobalt are easily decomposed, both by acids 

 and alkalies. For these reasons it does not seem to me tliat Blom- 

 s'rand's arrangement of the atoms in these salts deserves any special 

 preference at present. 



Blomstrand's views as to the constitution of the metal-jinunouias 



