270 



Dr. W. J. Russell. 



pound formed in each case was dried out of contact with air, and the 

 chlorine it contained determined ; the first one contained chlorine 

 corresponding to 4*85 per cent, of cobalt chloride, the second with 

 10*94 per cent., and the last with 28"02 per cent, of the cobalt chloride. 

 The compound having the formula 4CoOCoCl 2 , would contain 30'40 

 of cobalt chloride. Such are chemically the changes which occur. 

 Spectroscopically, the changes which the oxide undergoes are shown at 

 figs. 13, 14, and 15. Fig. 10 being the spectrum of the original blue 

 precipitate. 



These changes are, first, the fading of the band at 590, and then 

 the disappearance of all absorption on the less refrangible side of 590, 

 and ultimately the formation of a dark and well-defined band from 

 674 to 665, the faint band from 600 to 596 remaining. This change 

 has much the appearance of a division of this oxide band into two 

 bands of which the most refrangible part from 590 to 580 remains, 

 and the less refrangible part from 590 to 600 disappears. It seemed 

 possible that with a spectroscope of greater dispersive power to 

 separate the 690 band in the oxide into these two bands, but on trial 

 this did not take place. 



This spectrum, fig. 14, of the oxychloride, is in many respects similar 

 to the spectrum of second stage of the chloride in alcohol, although 

 certainly not identical with it, but in both cases the chloride appears 

 to be modified somewhat in the same way. That it is not an oxy- 

 chloride which is formed in the alcohol solution, seems proved by the 

 fact that as previously pointed out, this spectrum is formed by dis- 

 solving the chloride in acetic acid. 



Warm water decomposes this green compound, washing out the 

 cobalt chloride. This decomposition is very clearly shown in the ab- 

 sorption spectrum, if, for instance, some of the green compound be 

 prepared by precipitation as before described, and a little of it be 

 placed on a microscope slide under a covering glass, it gives very dis- 

 tinctly the spectrum shown at fig. 13. Now wash the precipitate on 

 a filter with warm water, and then, on taking another sample of it, 

 the two bands at 647 and 592 will be distinctly visible, that is, there 

 is evidently now oxide present. This oxychloride of cobalt may also 

 be formed in small quantities by digesting the precipitated oxide in a 

 solution either of nickel or potassium, or other chlorides, for on 

 examining the oxide thus treated, the change of the absorption spec- 

 trum is distinctly visible. 



The bromide of cobalt appears to react with the oxide in the same 

 way as the chloride. 



Generally the soluble cobalt salts of the oxygen acids do not give 

 band spectra, but only a broad absorption shading off on both sides 

 and extending from about 550 to 480 (fig. 16), and a considerable 

 amount of general absorption at the blue end of the spectrum ; for 



