192 



Profs. Li vein g and Dewar. 



[May 12, 



alternate diffuse triplets, including those near L and near S, much 

 expanded and reversed, and the group of five lines (2776 — 2782) 

 sometimes reversed. 



When the arc of a Siemens machine is taken in a magnesia crucible, 

 the strong line of the flame spectrum, wave-length 4570, is well seen 

 sharply defined ; it comes out strongly and a little expanded on drop- 

 ping in a fragment of magnesium. When a gentle stream of hydrogen 

 is led in through a hollow pole, this line is frequently reversed as ^a 

 .sharp black line on a continuous background. From comparing the 

 position of this line with those of the titanium lines in its neighbour- 

 hood, produced by putting some titanic oxide into the crucible, we 

 have little doubt that it is identical with the solar line 4570*9 of 



o 



Angstrom. 



When the arc is taken in a crucible into which the air has access, 

 it may be assumed that the atmosphere about the arc is a mixture of 

 nitrogen and carbonic oxide. When a stream of hydrogen is passed, 

 either through a perforated pole or by a separate opening, into the 

 erucible, the general effect is to shorten the length to which the arc 

 .can be drawn out, increase the relative intensity of the continuous 

 spectrum, and diminish the intensity of the metallic lines. Thus, with 

 & very gentle stream of hydrogen in a magnesia crucible, most of 

 the metallic lines, except the strongest and those of magnesium, dis- 

 appear. Those lines which remain are sometimes reversed ; those at 

 wave-length 2850 and the triplet near L being always so. With a 

 stronger stream the lines of magnesium also disappear, the b triplet 

 being the last in that neighbourhood to go, and \ and b 2 remaining 

 after & 4 has disappeared. 



Chlorine seems to have an opposite effect to hydrogen, generally 

 intensifying the metallic lines, at least those of the less volatile metals, 

 but it does not sensibly affect the spectrum of magnesium. Nitrous 

 oxide produces no marked effect ; coal-gas acts much as hydrogen. 



Spectrum of the Spark. 



In the spark of an induction coil taken between magnesium points 

 in air we get all the lines seen in the arc except two lines at wave- 

 lengths 4350 and 4166, two lines above U, and the series of triplets 

 more refrangible than the quintuple group about wave-length 2780. 

 The blue line wave-length about 4570 is seen in the spark without 

 a jar when the magnesium electrodes are close together, and the 

 rheotome made to work slowly, but requires for its detection a spec- 

 troscope in which the loss of light is small. 



On the other hand, some additional lines are seen. Of these, the 

 strong line at wave-length 4481 and the weak line at 4586 are well 

 known. Another faint line in the blue at wave-length 4808* has been 



* This line we first noticed in a former communication (" Proc. Roy. Soc," 



