chloride and
potassium ferrocyanide. For methods of preparation of the tetrammine
and triammine salts, see O. Dammer's _Handbuch der anorganischen
Chemie_, vol. 3 (containing a complete account of the preparation of
the cobaltammine salts). The diammine salts are prepared by the action
of alkaline nitrites on cobaltous salts in the presence of much
ammonium chloride or nitrate; they are yellow or brown crystalline
solids, not very soluble in cold water.

The above series of salts show striking differences in their behaviour
towards reagents; thus, aqueous solutions of the luteo chlorides are
strongly ionized, as is shown by their high electric conductivity; and
all their chlorine is precipitated on the addition of silver nitrate
solution. The aqueous solution, however, does not show the ordinary
reactions of cobalt or of ammonia, and so it is to be presumed that
the salt ionizes into [Co(NH3)6] and 3Cl'. The purpureo chloride has
only two-thirds of its chlorine precipitated on the addition of silver
nitrate, and the electric conductivity is much less than that of the
luteo chloride; again in the praseo-salts only one-third of the
chlorine is precipitated by silver nitrate, the conductivity again
falling; while in the triammine salts all ionization has disappeared.
For the constitution of these salts and of the "metal ammonia"
compounds generally, see A. Werner, _Zeit. fur anorg. Chemie_, 1893 et
seq., and _Berichte_, 1895, et seq.; and S. Jorgensen, _Zeit. fur
anorg. Chemie_, 1892 et seq.

The _oxycobaltammines_ are a series of compounds of the general type
[Co2O3.H2(NH3)10]X4 first observed by L. Gmelin, and subsequently
examined by E. Fremy, W. Gibbs and G. Vortmann (_Monatshefte fur
Chemie_, 1885, 6, p. 404). They result from the cobaltammines by the
direct taking up of oxygen and water. On heating, they decompose,
forming basic tetrammine salts.

The atomic weight of cobalt has been frequently determined, the
earlier results not being very concordant (see R. Schneider, _Pog.
Ann._, 1857, 101, p. 387; C. Marignac, _Arch. Phys. Nat._ [2], 1, p.
373; W. Gibbs, _Amer. Jour. Sci._ [2], 25, p. 483; J. B. Dumas, _Ann.
Chim. Phys._, 1859 [3], 55, p. 129; W. J. Russell, _Jour. Chem. Soc._,
1863, 16, p. 51). C. Winkler, by the analysis of the chloride, and by
the action of iodine on the metal, obtained the values 59.37 and
59.07, whilst W. Hempel an