550 Mr. C. W. Raffety on some Investigations of 



paper are reproduced from negatives obtained with the 

 following instruments : — 



1. Grating spectrograph of 22-inches focus. 



2. Grating spectrograph of 6-feet focus. 



3. Prismatic spectrograph with compound prism and 



camera of about 9-inches focus. 



Fig. 3, PL XIII., shows the hydrocarbon bands as they 

 appear, in the spectrum of the Meeker gas-flame, the com- 

 parison spectrum being that of iron in the carbon arc. The 

 line-series are seen to be complex, and the whole system of 

 bands is both beautiful find intricate. 



The region of " the three," and the less refrangible part 

 of the hydrocarbon spectrum as given by the are-in-flame, 

 is shown in fig. 4, PL XIII. The former are strong and 

 widened, and a close examination indicates a multiple 

 structure. 



In the case of the hydrocarbon bands the most striking 

 features, apart from the question of variation of relative 

 intensities, appear to be the new companion lines to the 

 isolated line of the flame spectrum about X 4324, which 

 appears single with small dispersion. 



The intricacy of structure of the bands, however, espe- 

 cially in. the " head " (X 4314), requires instruments of 

 considerable power, and only a partial resolution is shown 

 in these spectrograms. 



The ionization within the flame is so great, that to 

 maintain the disruptive discharge of the spark a source 

 not only of sufficient potential, but also of considerable 

 output is required to work with a condenser of the requisite 

 capacity, and a high-tension transformer was employed for 

 this purpose. 



In addition to an inductance in the primary circuit of the 

 transformer, another high-tension inductance was inserted 

 in the secondary circuit to control the current and prevent 

 arcing at the electrodes within the flame. 



In the oscillatory discharge circuit formed by the con- 

 denser and spark-gap, a variable inductance was introduced 

 so as to control the oscillation period of the spark. With 

 the spark between carbon electrodes in air, with capacity 

 but without self-induction, the line spectra of nitrogen and 

 carbon are seen. In the case of the same spark within 

 a flame of coal-gas there appear, in addition, the " Swan " 

 bands of carbon, the flutings of cyanogen, and, less strongly, 

 the hydrocarbon bands. 



The two spectra are thus superposed ; but their relative 



"I i 



