90 Professor Dewar [April 1, 



oxygen, and marsh-gas with oxygen, developed in general the same 

 lines as the hydrogen mixture, but gave a much brighter continuous 

 spectrum. Sulphuretted hydrogen, arseniuretted hydrogen, and 

 antimoniuretted hydrogen, exploded with oxygen, also gave very 

 bright continuous spectra, but no lines attributable to sulphur, 

 arsenic, or antimony. 



We have also tried explosions at higher pressures ; mixtures of 

 hydi'ogen, carbonic oxide, and marsh-gas respectively, with oxygen, 

 were compressed into the tube by a condensing syringe until the 

 pressure reached two and a half atmospheres, and in some cases three 

 and a half atmospheres. The general effect of increasing the pressure 

 was to strengthen very much the continuous spectrum, and also to 

 intensify the bright lines, so that photographs could be taken with a 

 smaller number of explosions. The lines previously observed to be 

 reversed were more strongly reversed, but no new lines which we can 

 attribute to the metals employed were noticed. No iron line more 

 refrangible than T showed itself in the photographs. But a banded 

 spectrum, of which traces had been noticed in the flash of the gases 

 at lower pressure, came out decidedly. This spectrum occupies the 

 region between P and E ; it is not a regularly channelled spectrum, 

 though probably under higher dispersion it would resolve itself into 

 groups of lines like the water-spectrum. In fact it seems to us most 

 probable that it is a development of the water spectrum, dependent 

 on the pressure. 



It seems very remarkable that metals so little volatile as iron, 

 nickel, and cobalt should develop so many lines* in the flash, while 

 more volatile metals show few or no lines. We do not know that 

 any lines attributed to the metals, as distinct from their compounds, 

 which have been observed in the gas-flame cannot be seen also in the 

 flash of the exploding gases, unless they be the blue lines of zinc 

 which Lecoq de Boisbaudran has seen faintly in the gas-flame when 

 zinc chloride was introduced. These are, however, so faint in the 

 flame, that they might easily escape notice in the much stronger 

 continuous spectrum of the flash. But iron, nickel, and cobalt show 

 no lines of those metals in a gas-flame. Mitscherlich (' Ann. de Phys. 

 u. Chem.' Bd. 121, St. 3), by mixing vapour of ferric chloride with the 

 hydrogen burnt in an oxyhydrogen-jet, obtained a number of the lines 

 of iron. These form three groups — one below D, one near E, and 

 one near G. The last two groups have a general correspondence 

 with the lines developed in the explosions in the visible part of the 

 spectrum ; but exact identification is not possible with his figure. 

 Of other metals he seems also to have found the same lines in the 

 oxyhydrogen-jet which we have seen in the explosions, but with 

 additional lines in several cases. Thus he found three zinc and as 

 many cadmium lines, two of mercury, four of copper, and so on. 



Gouy (' Comptes Rendus,' Ixxxiv. 1877, p. 232) has observed in 



* For detailed list of these lines see « Proc. Koy. Soc' vol. xxxvi. pp. 473-5. 



