OF ARTS AND SCIENCES. 45 



the other six hj chlorine, &c. Upon this view chloride of purpureo- 

 cobalt becomes (lONHo) = Cog 1 Clg, and chloride of luteocobalt 

 (12X113)= Co^ I Clg. In this manner the old theory of couplets or 

 conjugate ammonias may be rationalized and brought into harmony 

 with modern ideas. Two strong objections may be urged against the 

 theory here proposed. The first is that it requires us to consider an 

 atom of cobalt as hexatomic, while it exhibits in no other compound 

 an atomicity higher than four. The other objection has, I thinlv, 

 much greater weight. To explain upon this view the cases of isomer- 

 ism, to which I have myself directed attention, it becomes necessary to 

 assume that there are at least three allotropic forms of cobalt, — an 

 assumption wholly unsupported by any other and independent evi- 

 dence. On mature consideration I have therefore rejected this theory. 

 The objection which I have urged against my own view, that it re- 

 quires us to consider cobalt as hexatomic, with the atomic weigl)t 59, 

 may be avoided by considering the metal as tetratomic, and regarding 

 the two atoms as united by 6, 8, 10, or 12 atoms of ammonia, so that 

 the general formula of a normal cobaltamine chloride will be — 



CI CI 



1 I 



CI — Co — (nNH.,) — Co — CI 



I " I 



CI CI 



since n atoms of ammonia will always form a diatomic whole. The 

 other objection, that the theory obliges us to assume the existence of 

 several allotropic forms of cobalt, will, however, still remain. 



The view which I now adopt is in substance that of Blomstrand, 

 which affords, as I think, the simplest and most satisfactory explana- 

 tion of the whole series of ammonia-metallic compounds at present 

 known, and which, while not free from theoretic difficulties, is yet in 

 harmony with all the facts. 



According to this view, two atoms of tetratomic cobalt are asso- 

 ciated to form a hexatomic molecule, the six units of affinity being, in 

 all the cases at present known, in combination with four or six atoms 

 of ammonia, regarded as a diatomic. Six units of affinity remain, and 

 may be incompletely saturated by other atoms of ammonia or com- 

 pletely by chlorous elements or residues. Thus, on Blomstrand's view, 

 the chlorides of pur[)ureocobalt or roseocobalt and luteocobalt have 

 respectively the forumlas, — 



