and of Simple Substances. 173 



spectrum could only explain the occurrence of the feebler of 

 them. 



The compounds investigated by method 8 (liquid electrodes) 

 were more or less decomposed according to the magnitude of 

 the affinity of their constituents at this temperature, — which is 

 evident from the spectra ; the spectrum of chloride of copper is 

 seen very distinctly, while the spectra of bromide and of iodide of 

 copper are more difficult to recognize. 



Metallic copper investigated by method 7 (solid electrodes), in 

 chlorine, bromine, iodine, oxygen, and nitrogen, and in other 

 gases, always gives the spectrum of copper very distinctly, and, 

 further, that of the kind of gas, and in chlorine the spectrum of 

 chloride of copper very feebly. 



Hence the temperature produced by the passage of the spark 

 between solid electrodes is much higher than that between 

 liquids, because, in the first, iodide and bromide of copper remain 

 partially undecomposed, which in the latter are completely 

 decomposed. 



New lines occur in these spectra which neither belong to the 

 metal nor to the gas, and whose formation and origin I have not 

 yet investigated ; for the lines are mostly so obscure that a cer- 

 tain result could not be deduced from them. Sulphate and 

 nitrate of copper, and iodide, bromide, and chloride of copper, 

 examined in the different gases as solid salts by method 7, behave 

 like metallic copper. 



Bismuth in the formation of its spectrum exhibits almost 

 entirely the same phenomena as copper, excepting that the com- 

 pounds are much more easily decomposed, when the spectrum of 

 the metal appears. The spectra of chloride, bromide, and iodide 

 of bismuth are obtained by method 6 (volatilization in the glass 

 tube). But if the tube is strongly heated, the compounds are 

 partially decomposed, even at this temperature, in the manner 

 described. The spectrum of an oxide of bismuth readily occurs 

 along with that of the haloid salt when water is present in the 

 latter or is formed in appreciable quantity. 



The spectrum of the metal is best obtained if any bismuth 

 compound or the metal itself is brought into the reducing flame 

 in method 2 (hydrogen, or coal-gas, and 0) ; that of the oxide, 

 if in the same method bismuth is held in the strongly oxidizing 

 flame. 



In whatever way the spectrum of metallic bismuth is obtained, 

 whether by electricity or by flame, it is under all circumstances 

 the same. 



The compounds of calcium, of strontium, and of barium give 

 almost as many spectra as do those of copper and bismuth. But 

 to obtain them pure is more difficult than with the above metals. 

 The spectra of barium and of strontium by the flame, I could 



