118 Profs. G. D. Liveing and J. Dewar. On the [June 1(5, 



chemical changes, and deriving their energy directly from the 

 chemical energy of the interacting substances. When the absorption 

 spectrum of a compound is observed to correspond with its emission 

 spectrum in a flame, we may infer that the vibrations are those which 

 the compound molecule assumes when sufficiently heated. But there 

 are not many cases in which this has been observed. We have 

 observed it in the case of cyanogen (' Eoy. Soc. Proc.,' vol. 44, p. 247, 

 note), but we are not certain of any other case. The difference 

 between the spectrum of the base of a flame and that of the upper 

 part, observed in many flames, lends support to the supposition that 

 there are rays which originate in the chemical change, perhaps 

 occurring in the molecules which are in intermediate stages of the 

 change, and not assumed by the molecules which are the final pro- 

 duct, even when intensely heated in the upper part of the flame. 

 The fact that the same rays which are seen in the base of a flame 

 may be sometimes generated by electric discharges in the gases which 

 are burnt in the flame, or in their products of combustion, is not at 

 all inconsistent with this supposition, for such discharges certainly 

 have electrolytic effects, and may very well give rise to molecules in 

 the intermediate stages between one state of chemical combination 

 and another.* The flames of substances, such as the organo- metallic 

 compounds, into which metals enter as chemical ingredients, have not 

 hitherto, so far as we know, been observed, and it is to two such 

 flames that these notes refer. 



Spectrum of the Flame of Nickel- Carbonyl. 



The remarkable compound of nickel and carbonic oxide, Ni(CO) 4 , 

 discovered by Mr. Mond, burns in air with a luminous, smoky flame, 

 and the spectrum it emits appears to be a continuous one. When the 

 vapour is burnt in oxygen instead of in atmospheric air, the spectrum 

 still appears to be quite continuous ; in fact, such a spectrum as 

 carbonic oxide, without any nickel, gives under similar circumstances. 

 This, however, is only in appearance, because the brightness of the 

 continuous spectrum overpowers the feebler bands and lines which 

 belong to the flame of the nickel compound. These bands and lines 

 come out when the vapour of the nickel compound is diluted with a 



* It is sometimes assumed in books on chemistry that the atoms which form a 

 chemical compound can never be in an intermediate state between complete separa- 

 tion and complete combination. So inconceivable an assumption would hardly have 

 been made except to support a theory, but it has nevertheless obtained a certain 

 currency. It is supported by no fact and no analogy. Two atoms which are 

 within the spheres of each other's influence, but have not yet reached the state of 

 relative tranquillity which we recognise as chemical combination, may very con- 

 ceivably be the seat of very violent agitation and vibratory motions, which cease 

 when they are actually combined. 



