On the Absorption Spectra of Cobalt Salts. 261 



decomposition. On examining this fused mixture it is found to give 

 an absorption spectrum (shown at fig. 3) quite different from that 

 given by the chloride alone. On varying the proportions of the two 

 salts, even taking as much as twenty molecules of the potassium salt 

 to one of the cobalt salt, still the spectrum remains the same ; in fact, 

 if only a trace of the fused or hydrated cobalt chloride be added to 

 potassium chloride, the fused mass gives this same spectrum. On 

 increasing the thickness of the layer of salt examined or the amount 

 of cobalt present in a layer of constant thickness, the absorption 

 extends between the bands, and quickly reaches 600. Now, when the 

 absorption has reached thus far, a faint band from 540 to 535 becomes 

 visible, a band agreeing in position and character with the one pro- 

 duced by the pure chloride ; increase the absorption still more, both 

 sides close in, leaving only a band of light from 580 to 550. On 

 making a similar experiment with the bromide of cobalt, fusing it with 

 potassium bromide, an absorption spectrum is obtained (given at 

 fig. 4), and if the iodide of cobalt and potassium iodide be fused to- 

 gether, a spectrum (shown at fig. 5) is obtained ; the first two spectra 

 are strikingly alike. In the above order the absorption bands are 

 nearer and nearer to the red end of the spectrum, and only one band 

 of the iodide is clearly visible, still one cannot but suspect the existence 

 of the others, although not visible. With a considerable thickness of 

 the bromide a band from 573 to 558 is very visible, the analogy of the 

 532 band in the chloride. Fluoride of cobalt (for a specimen of which 

 I have to thank Dr. Gore), neither alone nor with potassium fluoride, 

 gave any visible absorption bands. 



In place of potassium chloride, sodium chloride, ammonium chlo- 

 ride, zinc chloride were used ; they all, when cold, gave the same 

 absorption spectrum, but with this peculiarity, that the exact position 

 of the two bands between about 600 and 625 varied with each salt to 

 a small extent. With the potassium chloride the bands were nearest 

 to the red, and with the zinc chloride nearest to the blue. 



The distance between these two bands appears to remain nearly if 

 not quite constant. With regard to the absorption which extends 

 from 650 to 700, if only a thin layer of the salt be used, it will be 

 clearly seen that it is resolvable into a dark band from 696 to 675, and 

 a fainter band at 656 (figs. 3 and 6). The action of heat on these 

 bodies is curious and interesting, the absorption at the red end of the 

 spectrum is intensified, the bands disappear in a mass of dark absorp- 

 tion, and with regard to the two bands in the orange they gradually 

 merge into a single band ; and on still further increasing the heat 

 (but at a temperature still considerably below the fusing point of the 

 mixture) the band at 625 disappears, and there is a broad band beginning 

 sharply about 620, and extending nearly up to 600. On cooling, pre- 

 cisely the same changes are gone through in reverse order. On re- 



