1887.] on Light as mi Analytic Agent. 91 



the inner green cone of a modified Bunsen burner fed with gas mixed 

 with spray of iron-salts, four green lines of iron which we did not fi.nd 

 in the flash. He saw two of the blue lines, but not the other lines 

 which we have noticed. In like manner with cobalt, he observed two 

 feeble blue rays which we did not see in the exj)losions ; also one zinc, 

 one cadmium, and one silver line which we did not see ; and he did 

 not notice the green copper line which we always have seen in the 

 explosions. In other cases he has noticed the same lines that we 

 have noticed. 



Comparing the spectrum of the explosions with that of iron wire 

 burnt in a jet of coal-gas fed with oxygen, they may be called 

 identical. We find in them generally the same lines and the same 

 relative strengths of the lines. For instance, in the explosion- 

 spectrum the strength of the groups of lines on either side of M and 

 the line at wave-length 3859*2 is decidedly greater as compared 

 with the other lines than it is in the arc-spectrum of iron. It is the 

 same in that of iron burnt as above mentioned. T, however, comes 

 out more strongly in the last-mentioned spectrum than in the ex- 

 plosions. 



German-silver wire burnt in the coal'gas and oxygen jet gave the 

 same nickel and copper lines as were developed in the explosions. 

 Silver wire gave in the same jet the two silver lines near P, but no 

 channelled spectrum. SjDray of cobalt chloride gave also the same 

 lines as in the explosions, with a few additional; while spray of 

 manganese chloride gave the strong manganese triplet at wave-length 

 about 2800, more refrangible than anything observed in the explo- 

 sions, besides the usual violet triplet. 



On the whole the spectra produced by the jet of coal-gas and 

 oxygen are very similar to those of the explosions as far as the 

 metallic lines go ; they exhibit a few more lines, or it may be these 

 are more easily observed. 



Of the green and blue lines of iron seen by us in explosions nine 

 are registered by Watts as occurring in the flame of a Bessemer con- 

 verter ; or at least the lines he gives are so near that we cannot 

 doubt their identity. 



When we come to make a comparison with the spectrum of the 

 spark-discharge from a solution of ferric chloride, the differences 

 become more marked. Not only are there many more lines in the 

 spark-spectrum, but the relative intensities of those lines which are 

 common to both spark and explosion are very different, and two of 

 the iron lines seen in the explosions appear to be absent from the 

 spark. The differences between the spectrum of the spark taken 

 from a liquid electrode and that given by solid electrodes has usually 

 been attributed to the lower temperature of the former ; but the 

 absence from the former spectrum of the line at wave-length 4132, 

 and the feebleness of the line at wave-length 4143, both strong lines 

 in the arc and in the explosions, as well as in the spark between 

 solid electrodes, seem to indicate that the differences of spark-spectra 



