14 On Double Spe&ira. [January, 



therefore be less than 2000 C. It is also given (with the 

 addition of the groups £ and 6) by the flame of cyanogen in 

 air, the temperature of which we have seen to be 3297 , and 

 by the flame of cyanogen in oxygen, which is the hottest 

 flame known, and is probably at least 5000 or 6000 C. We 

 see thus that the lines of the first spectrum are given by 

 carbon heated (at any rate) between 2000 and3000° C. Now 

 it can be shown that the second carbon spectrum can also 

 be produced by carbon heated within the same range. 



This second spectrum is not obtainable from any flame at 

 all, but is produced by the electric discharge in either car- 

 bonic oxide, or olefiant gas, or carbon disulphide. 



We have seen that the sodium spectrum contains, when 

 the temperature is sufficiently high, besides the well-known 

 double line, four other lines, each of them double ; Na/3 at 

 56, Na7 at 75*5, NaS at 83*2, and Nae at 43. In the Bunsen 

 flame, the D lines only are obtained ; but, if the tempera- 

 ture of the flame be increased, these other sodium lines 

 come out one by one, and in the order given above, Na/3 

 becoming visible almost precisely at the temperature at 

 which platinum melts — that is, 2000 C. ; so that, if a bead 

 of sodium carbonate be brought into any flame incapable of 

 fusing platinum, only the D lines will be seen, but, if the 

 flame be hot enough to fuse platinum, Na/3 will also be 

 visible. 



For example — 



The Bunsen flame gives the D lines only, and is in- 

 capable of fusing platinum. 



The flame of coal-gas, fed by a jet of air mixed with a 

 little oxygen, gave Na/3, but not Nay. The flame fused 

 platinum easily. 



The flame of carbonic oxide in air (temperature, 1997° C.) 

 gave only D ; it is incapable of fusing platinum. Carbonic 

 oxide fed by oxygen (temperature, 3033 C.) gives Na/3 and 

 Nay. 



We conclude, therefore, that Na/3 indicates a temperature 

 of at least 2000 C, and that Nay comes out about 3000 C. 



If now the discharge of an induction coil be passed 

 through a Geissler's tube containing carbonic oxide into 

 which some pieces of sodium have been introduced, the carbon 

 spectrum No. 2 is obtained ; and if the sodium be volatilised, 

 the sodium lines come out one by one as the tube gets hot, 

 and the carbon lines are seen simultaneously with Na/3, and 

 (at first) without Nay. Hence we conclude that this 

 second spectrum is also produced by carbon between the 

 temperatures of 2000 and 3000 C, and we are therefore 



