246 



Profs. Gr. D. Liveing and J. Dewar. [May 31, 



L. As we had not observed these features in the spectrum of the 

 spark or arc, and could not trace their connexion with any compound, 

 we concluded that they were produced by magnesium only at the 

 comparatively low temperature of the flame. We have since found 

 that they are not produced by the metal at that temperature only, but 

 are exhibited as strongly, or even more strongly, in the arc between 

 electrodes of magnesium. In the latter case they appear concurrently 

 with the line at 448 L and other lines which seem to belong to high 

 temperatures. We must therefore regard them as not only produced 

 at the temperature of flames, but as persistent at temperatures very 

 much higher. 



The different circumstances under which we have observed this 

 triplet are as follows : — 



In the oxyhydrogen flame when a very small piece of magnesia is 

 held in it. In this case the outer two lines of the triplet are much 

 stronger than the middle line (X3724 about), which in some of onr 

 photographs does not show at all. It should be noticed that the least 

 refrangible of the three lines (\3730 about) is in general more diffuse 

 and not quite so bright as the two more refrangible lines. Magnesia 

 in the oxyhydrogen flame also gives rise to some bands close to and 

 more refrangible than the triplet, and to another still more refrangible 

 bnt less bright triplet, in which the lines are set at nearly equal 

 distances from each other, with the approximate wave-lengths 3633' 7, 

 3626" 2, 36206. These additional bands and triplets are not really 

 absent from the flame spectrum, for traces of them may be seen in 

 some of our photographs of the magnesium flame, but they seem 

 relatively brighter in the oxyhydrogen flame with magnesia, and the 

 longer exposure of the photographic plate in the latter case helped to 

 bring them out. They seem to come out more strongly under the 

 conditions which make both the green bands of the oxide and the 

 b group show well. 



The triplet near M is also produced when magnesium oxychloride 

 and when magnesium chloride is substituted for magnesia in the oxy- 

 hydrogen flame, and in the former case the more refrangible triplet is 

 developed as well. 



When carbonic oxide and oxygen are substituted for hydrogen and 

 oxygen, both triplets are developed in the part of the flame near the 

 magnesia, and in this flame the middle line of the triplet near M 

 (\3724 about) is as strong as it is in the flame of burning magne- 

 sium. 



The proper adjustment of the thread of magnesia in this flame is a 

 much more delicate matter than in the oxyhydrogen flame. In fact, 

 we made many experiments which were failures before we succeeded 

 in getting satisfactory results ; and latterly, iu order to be certain of 

 success, we had to fill a gas-holder with a mixture of carbonic oxide 



