162 MR. CHARLES DE WATTEVILLE ON FLAME SPECTRA. 



spectrum are ever seen in the flame spectrum. The resemblance, however, is very 

 marked between the flame spectrum and that of the spark, in which the oscillatory 

 character is controlled by the introduction of a self-induction into the discharging 

 circuit of a condenser, and which for the sake of brevity may be called an oscillatory 

 spark. In the latter case, as is well known from the work of Dr. HEMSALECH, the 

 spark spectrum is considerably simplified. Moreover, although the flame spectrum 

 will contain only the lines which belong to the spectrum of the oscillatory spark, yet 

 all the lines of the latter will not be found in the flame spectrum,- the missing lines 

 l>eing those which are peculiar to the ordinary spark spectrum, and which only exist 

 in the immediate neighlxnirhood of the electrodes, becoming shorter and shorter, and 

 finally -disappearing as the self-induction is increased. 



The preceding paragraph refers to metals other than those belonging to the iron 

 group. On the contrary, there is a most striking similarity between the flame spectra 

 of iron, of nickel and of cobalt and the oscillatory spark spectra of the same metals in 

 the region included between about 4300 and 2700 Angstrom units. The similarity of 

 the two spectra is so great that, except for very small differences of intensity, the 

 oscillatory spark spectrum, which is photographed as a comparison spectrum in the 

 centre of the flame s]>ectrum, appears to be a prolongation of the latter. This may 

 be seen in the photographs placed at the end of this paper. 



It should be noticed that if in the visible portion of the spectrum certain lines 

 appear to be missing, it is probably because the continuous spectrum which appears 

 in this region of the spectrum prevents these feeble lines from being seen. This may 

 explain why M. GOUY was not able to observe the nickel lines which are found on 

 the photographs taken with various salts of nickel, viz., the sulphate, chloride, and 

 ammoniacal chloride. In the ultra-violet the spectrum of the flame appears to fade 

 away a little more rapidly than that of the oscillatory spark, but it is probable that 

 this difference would be reduced by prolonging the time of exposure ; since it is, of 

 course, the radiations of the shortest wave-length which are most absorbed by 

 different media. 



What are the reasons for this similarity between the spectrum of the flame and 

 the spectrum of the oscillatory spark ? Until we have conclusive evidence to the 

 contrary, we must assume that it is a question of temperature. On the one hand, 

 the increase in the number of lines of the flame spectrum obtained by the use of the 

 sprayer may be attributed to the fact that the hottest regions of the flame take part 

 in the production of the phenomena, and, on the other hand, the diminution in the 

 number of lines in the spark spectrum when the spark becomes oscillatory is due to a 

 diminution of its temperature. In fact, in the case of the ordinary spark, the 

 whole of the energy liberated in the secondary circuit, of the induction coil by the 

 breaking of the current in the primary circuit is dissipated in a single rush ; whereas, 

 in the case of the oscillatory spark, the rate of dissipation is greatly reduced by the 

 conditions under which the discharge takes place. It is really the temperature at 



