1881.] Investigations on the Spectrum of Magnesium. 197 



seen in the arc, and may be reasonably ascribed to the higher tem- 

 perature of the spark, may be seen in the spark at the same time 

 as the line at 5210 when hydrogen is present. Nevertheless, tem- 

 perature does seem to affect the result in some degree, for when a 

 large Leyden jar is used, and the gas is at' the atmospheric pressure, 

 the line almost disappears from the spark, to reappear when the 

 pressure is reduced ; but by no variation of temperature have we been 

 able to see the line when hydrogen was carefully excluded. 



"A line of the same wave-length has been seen by Young in the 

 chromosphere once. Its absence from the Fraunhofer lines leads to 

 the inference that the temperature of the sun is too high (unless at 

 special times and places) for its production. If it be not due to a 

 compound of magnesium with hydrogen, at any rate it occurs with 

 special facility in the presence of hydrogen, and ought to occur in the 

 sun if the temperature were not too high. 



" We have been careful to ascribe this line and its attendant series 

 to a mixture of magnesium and hydrogen rather than to a chemical 

 compound, because this expresses the facts, and we have . not yet 

 obtained any independent evidence of the existence of any chemical 

 compound of those elements." 



Fig. 4 shows more completely than we have gi^en it before the 

 general character of this spectrum, which consists of two sets of 

 flutings and a pair of fainter bands, the flutings closely resembling 

 in character the hydrocarbon flutings, each fluting consisting of 

 a multitude of fine lines closely set on the less refrangible side 

 and becoming wider apart and weaker towards, the more refrangible 

 side, but extending under favourable circumstances much further 

 than is shown in the figure. The set in the green is the stronger, 

 and it was to this that our former observations were confined. It 

 has two flutings, one beginning at about wave-length 5210 and the 

 other close to b Y on its more refrangible side. The other set consists 

 of three principal flutings, of which the first begins at about wave- 

 length 5618, the next at about wave-length 5566, and the third begins 

 with three strong lines at about the wave-lengths 5513, 5512, 5511. 

 Both sets are very well seen when a magnesium wire is burnt in the 

 edge of a hydrogen flame, and in the arc in a crucible of magnesia 

 when a gentle current of hydrogen is led into it. The less refrangible 

 edges of the bands are at wave-lengths about 4849 and 4803. 



As Mr. Lockyer, in a paper entitled " A New Method of Spectrum 

 Observation" ("Proc. Roy. Soc," vol. 30, p. 22) has brought for- 

 ward this spectrum as illustrative and confirmatory of his views 

 regarding the possibility of elemental dissociation at different heat- 

 levels, we have been induced to review our former work. The view 

 taken by Mr. Lockyer may be expressed in his own words. 



" The flame spectrum of magnesium perhaps presents us best with 



