1887.] on Light as an Analytic Agent. 93 



Eecently published observations by Bertbelot and Vieille (' Comptes 

 Eendus,' xcviii. 1884, p. 548) put the pressure of explosion of 

 oxygen and hydrogen at 9 • 8 atmospheres, and of carbonic oxide and 

 oxygen at 10*1, and the corresponding temperatures 3240^ and 3334°. 

 The pressures determined by the two observers agree closely, and the 

 calculated temperatures are not very discordant. On the whole, we 

 cannot be wrong in assuming the temperature of the exploding gases 

 to be about 3000° ; and we see that at this degree such metals as iron, 

 nickel, and cobalt are vaporous, and emit many characteristic rays, 

 and that by far the greatest part of these rays lie between narrow 

 limits of refrangibility, G and P. Even for other metals there is a 

 predominance of rays in the same part of the spectrum. The lines 

 of lead, potassium, and manganese, three out of four lines of thallium, 

 and two-thirds of those of chromium, observed in the explosions, fall 

 within the same region. It must not be inferred that these facts 

 indicate the limit of the rate of oscillation which can be set up in 

 consequence of an elevation of temperature to 3000^, because we know 

 that the spectrum of the lime-light extends much further. But it 

 might be possible to establish a sort of spectroscopic scale of tempera- 

 tures if the lines which are successively developed as the temperature 

 rises were carefully noted. Thus the appearance of the iron line T 

 seems to synchronise with temperature of about 3000°. The lithium 

 blue line is invisible in the flame of an ordinary Bunsen burner, but 

 is just visible at the temperature of the inner green cone formed by 

 reducing the proportion of gas to air in such a burner, while in the 

 exploding gas the green line too is seen. It seems to need a tempera- 

 ture above 3000° to get the aluminium lines at H. Probably no line 

 is ever abruptly brought out at a particular temperature — it will 

 always be gradually developed as the temperature rises; yet the 

 development may be rapid enough to give an indication which may 

 be useful in default of means of more exact measurement. In former 

 papers treating of spectroscopic problems (' Proc. Eoy. Soc' vol. xxxiv. 

 p. 130, and xxix. p. 489) we have more than once adverted to the 

 necessity of the study of the spectra both of flames and of the electric 

 discharge under modified conditions of pressure. The projected ex- 

 periments on the arc in lime-crucibles have not yet been carried out ; 

 but the present is a first instalment of a study of flame-spectra under 

 such conditions. 



[J. D.] 



