92 Hartituy—An Investigation of the Connexion between 
magnesium chloride in a Delachanel and Mermet’s tube ; it is also at the head of 
the magnesium-hydrogen band. 
The lines of the triplet are common to are, spark, and the flame of burning 
magnesium. 
The experiments of Liveing and Dewar have proved beyond all question that 
magnesium oxide is reduced by three different reactions: in the oxyhydrogen 
flame, in the flame of cyanogen when burnt in oxygen, and of carbon monoxide 
mixed with half its volume of oxygen, and burnt as it issues from the gas-holder 
(Proc. Roy. Soc., 1888, 44, 241). The bands are called the ‘‘ oxide bands,” 
because, as Liveing and Dewar pointed out in their second paper (1888), p. 242, 
‘they are due either to magnesia, or to the chemical action of oxidation,” and 
again on p. 248, ‘‘the triplet near J and its associated bands” are due, ‘‘ not to 
merely heated magnesium, but to the oxide, or to vibrations set up by the process 
of oxidation.” 
The same remark, based upon similar experimental evidence, applies to 
‘calcium ovide” bands, also to the bands observed in the oxyhydrogen flame- 
spectra of strontium and barium oxides,* that is to say, they are due to vibrations 
set up probably by the chemical process of oxidation. ‘To return to magnesium, 
it has been stated by Olmsted that he has found no strong band corresponding to 
that observed by me in 1894 in the magnesium sulphate spectrum, between the 
* It is necessary that this should be clearly stated, because the term has not been applied in this 
sense by recent investigators ; thus, de Watteville remarks on the occurrence of bands with magnesium, 
barium, strontium, calcium, copper, tin, and manganese: ‘‘The study of these bands, and the nature of 
the compound to which they are due, forms no part of the present investigation.” 
Since the publication of my paper ‘‘On the Thermo-Chemistry of Flame-Spectra at High Temperatures ”’ 
(Proc. Roy. Soc., A, vol. 79, 1907), two communications have been made upon the subject of Bunsen 
flame-spectra by Carl Fredenhagen (Phys. Zeitschr., 1907, 8, 407-415, and Ber. 1907, 40, 2858-2861). 
They are considered by the author to be the result of chemical reactions, and not solely the effect of a 
high temperature, as I have already shown. The argument employed is, that they are always obtained in 
oxygen flames, but not in the hydrogen-chlorine flame; hence the spectrum reaction is due to the union 
of the metals with oxygen. In support of this contention, Smithell’s modified Bunsen burner (Chem. 
Soc. Trans., 1982, 61, 204) was fed simultaneously with lithium and copper salts, by employing 
Beckmann’s method, which is in effect that of E. Cleminshaw (Phil. Mag. [5], 19, 365-368, 1885). It is 
found that the outer zone of the flame is coloured green, while the inner zone is coloured red. This 
is accounted for by the copper being oxidized in the outer zone, but not in the inner; but the lithium 
haying a greater affinity for oxygen than hydrogen, is oxidized in the inner zone. The proper 
interpretation may be exactly the opposite, the copper oxide being reduced in the outer zone, and the 
lithium only at the higher temperature of the inner zone; but, as de Watteville has shown, both series of 
lines may appear in the green flame of the inner zone. 
It is evident that the proofs I had given, of the manganese bands belonging to the element, and not to 
its oxide, had been overlooked by de Watteville. Also C. M. Olmsted, ‘‘ Die Bandenspektra nahe 
verwandter Verbindungen,” Zeitschr. f. wissenschaft, Photographie, 1906, 4, 255, refers to the bands 
of MgO, CaO, &c. 
