of Liyht Radiations emitted by Iron Vapour. 7 



There is no advantage in using more than 6 flames one 

 © © 



behind the other, for, as Gouy has shown, a flame charged 

 with radiating vapour absorbs about 50 per cent, of a similar 

 flame placed behind it. 



The gases in the air-coal gas mixture were so proportioned 

 as to produce blue-coloured cones; it was found that the 

 light emission by the metal vapour was at a maximum in this 

 case. The heights of the cones given by burner No. 2 were 

 from 3 to 8 millimetres, according to the velocity of the gas 

 mixture. 



Burners for oxy-coal gas and oxy-liydrogen flames. 

 The burner selected was one in which the two gases are 

 mixed before being burnt. The oxygen and coal-gas or 

 hydrogen arrive through separate channels at the base of a 

 small mixing-chamber, and thence pass through a short tube, 

 about 2 in. long, to the nozzle from which the mixture 

 escapes through an orifice of 1 millim. diameter. 



Burner for oxy-acetylene flame. 

 This was one of the ordinary industrial type and not 

 really suitable for spectroscopic work, but no other pattern 

 was available at the time. The acetylene passes into the 

 mixing-chamber at low pressure. The oxygen, on the other 

 hand, is forced under pressure through a very fine orifice 

 into the mixing-chamber and diffuses into the acetylene. 

 The mixture then passes through the nozzle and a 1 millim. 

 orifice. It was, of course, impossible to charge the oxygen 

 with the material from a sprayer, because the fine orifice 

 through which the oxygen has to pass becomes immediately 

 clogged. So the spray from a Grouy apparatus was passed 

 into the outer parts of the flame by means of a jacket fixed 

 round the nozzle of the burner. This is a bad method of 

 obtaining the spectrum of a flame, for the material does not 

 pass through the cone region and I adopted it, though most 

 reluctantly, only because previous experience with the oxy- 

 acetylene flame had shown me that iron vapour does not 

 emit spectrum radiations in the inner cone of this flame. In 

 the case of nickel and cobalt this method failed altogether. 



Spectrog7*apli and projection apparatus. 



A serviceable spectrograph was constructed out of the 



remains of a Kirchhoff and Bunsen four-prism spectroscope*. 



Only three prisms were available, each having a refracting 



angle of 45°. The objective of the observing telescope was 



* The orio-inal instrument is described in Roscoe-Schusters ' Spectrum 

 Analysis,' p. 70 (1890). 



