On the Absorption Spectra of Cobalt Salts. 265 



each of the solvents, and that this banded spectrum was really that 

 of a hydrate, and not of the chloride. To, settle this point the follow- 

 ing experiments were carefully carried out : Some of the mixed 

 chloride of cobalt and potassium was dried and fused in a glass tube 

 in a slow current of very thoroughly-dried carbon dioxide, in the first 

 instance for twenty minutes ; it was then allowed to cool and was 

 spectroscopically examined, and found to be quite unchanged ; again 

 the heating and the slow passage of dry gas was continued for forty 

 minutes, and afterwards for two hours, but in no case was the least 

 change apparent, the spectrum-giving substance was evidently as 

 abundant after all the heating as before, and no trace of any other 

 spectrum was visible. The experiment was also made of fusing 

 separately in different parts of same tube, filled with dry carbon 

 dioxide, the cobalt chloride, and potassium chloride, and, after they had 

 been in a state of fusion for some time, allowing them partially to 

 mix, and sealing up the tube; where the two had come into contact 

 this spectrum was visible and nowhere else. The hydrochloric acid 

 solution of the cobalt chloride in one respect, however, is different 

 from any of the other solutions, for, as before mentioned, only one 

 spectrum is obtainable from it, whereas with the other solvents, the 

 spectrum varies with the amount of chloride in the solution. Thus 

 a very dilute alcoholic solution gives a spectrum (fig. 7) evidently 

 different from the former one given by the saturated solution. To 

 obtain clearly this second spectrum dry alcohol must be used, and 

 should contain about 0*008 grm. of cobalt chloride in 100 cub. centims. 

 of the alcohol ; this amount of the cobalt salt gives only a faint purple 

 colour to the alcohol, and requires a solution of about 2-J- inches in depth 

 to show the spectrum. If the strength of the solution be gradually 

 increased so that it becomes of a strong blue colour, and contains 

 about 20 grms. of the chloride in 100 cub. centims. of alcohol, and 

 requiring for observation only about one-eighth of an inch thickness 

 of the solution, a third spectrum (fig. 8) is obtained. This, however, 

 appears in part to be a compound spectrum, formed by the presence 

 in the solution of the bodies which produce the former spectra. In- 

 dications of two bands, 610 and 625, for instance, are visible, but 

 what is characteristic of this spectrum, and does not occur when the 

 solution is very strong or very weak, is the band at 585. In order to 

 determine whether any other changes would occur when the solution 

 was still more dilute, some of it requiring a depth of 2 inches to show 

 the spectrum was diluted to such an extent that 6 feet of the liquid 

 was required. This great dilution produced no change whatever in 

 the spectrum. Other liquids, such as dry amylic alcohol or amylic 

 acetate, which readily dissolve cobalt chloride, give spectra cor- 

 responding to those given by ethylic alcohol. With methylic alcohol, 

 however, the first or most dilute stage was not obtained, but this 



t 2 



