1888.] Investigations on the Spectrum of Magnesium. 243 



the bands in the absence of oxygen, and their increased brilliance in 

 that gas, leave little room for doubt that they are due to the oxide, or 

 to the process of oxidation. It may be assumed that at a sufficiently 

 high temperature magnesia will be decomposed, but magnesia is a 

 very stable compound, a great amount of heat is developed in its 

 formation, and it probably requires a temperature far above that of 

 burning magnesium for its complete dissociation. This is consistent 

 with the appearance of the bands in the spectrum of the flame of the 

 burning metal, as well as in the condensed spark when the other 

 conditions are favourable for the formation of the oxide, or for its 

 stability when formed. In our earlier observations, we obtained in the 

 visible region nothing but a continuous spectrum from magnesia 

 heated with the oxyhydrogen blowpipe ; neither the b group, nor 

 X4570, nor the triplet near L appeared, but at the same time X2852 

 was not only strong, but was strongly reversed. We now find that 

 this result, so far as it was negative, was a consequence of using too 

 large a mass of magnesia to be adequately heated by the flame. If 

 the piece of magnesia is very small, such as a fragment of the ash of 

 burnt magnesium ribbon, most of the spectrum of burning 

 magnesium is developed in the flame for a short distance from 

 the piece of magnesia. It was not very easy to make these experi- 

 ments successfully. About 3 inches of magnesium ribbon were 

 burnt in air, and the ash carefully heated in the upper part of the 

 oxyhydrogen flame to render it dense. The thread of magnesia so 

 obtained was held horizontally with its end projecting into the oxy- 

 hydrogen flame so as to approach the boundary of the inner cone, and 

 if the current of gas were not too strong all that was further 

 necessary was to move up the thread horizontally as the end was 

 worn away. When the magnesia was placed as described, the whole 

 upper part of the flame was of a fine azure-blue colour. Under these 

 circumstances, the flame shows the b group and the magnesium- 

 hydrogen series close to it, the bands in the green, the triplet near L, 

 the triplet near M of the flame of burning magnesium, with the 

 group of bands in that region, and the line V2852. It is remarkable 

 that the proportions in which the oxygen and hydrogen are mixed 

 affect the relative intensities of different parts of the spectrum. In 

 general, both the metallic lines of the b group and the bands of the 

 oxide are easily seen ; but if the oxygen be in excess the bands of the 

 oxide come out with increased brightness, while the b group fades or 

 sometimes becomes invisible. On the other hand, if the hydrogen be 

 in excess the bands fade, and the b group shows increased brilliance. 

 There can hardly be much difference in the temperature of the flame 

 according as one gas or the other is in excess, but the excess of 

 oxygen is favourable to the formation and stability of the oxide, 

 while excess of hydrogen facilitates the reduction of magnesium and 

 vol. xuv. s 



