120 Profs. G. D. Liveing and J. Dewar. On the [June 16, 



discharges in carbonic oxide at low pressure. The positions of the 

 bright edges of the bands in the flame of the nickel- carbon yl have 

 some correspondence with those of the bands produced by electric 

 discharges in carbonic oxide, but it is not a. very close one, and may 

 be only accidental. With the dispersion employed, which gives a 

 difference of deviation of 2 52' between D and F, there was no sign 

 of a resolution of these bands into lines. The brightest bands had, 

 however, their more refrangible edges pretty sharply denned, while 

 the less bright bands, especially those in the blue, were very hazy. 

 A certain amount of continuous spectrum, of course, overlay the 

 bands, and made them somewhat less distinct. Photographs show 

 that this continuous spectrum continues as far as \3500, bat fading 

 sensibly from A, 4200 onwards. The photographs do not show any 

 extension of the bands beyond the blue. 



Besides the bands, a few lines, but only a few, in the visible part of 

 the spectrum, extend into the upper part of the flame. Of these few- 

 only one is a known line of nickel ; it is the green line X 5476. This 

 was also the only line of nickel which we observed in the visible part 

 of the spectrum in explosions of hydrogen and oxygen in a nickel- 

 lined tube ( Boy. Soc. Proc.,' vol. 36, p. 475). 



In the ultra-violet part of the spectrum of the flame a great 

 number of nickel lines were photographed ; indeed, by far the greater 

 part of the lines of nickel found by us in the arc (' Phil. Trans.,' 

 vol. 179 (1888), A, p. 247) between X3972 and \2943'5. In this 

 case also there is a close correspondence between the spectra of the 

 flame and of the explosions, except that the lines of the flame are 

 much more numerous than those recorded of the explosions. This 

 difference, however, is probably due to the much shorter exposure of 

 the photographs of explosions. Although the photographs show 

 lines as high as \ 2943, the lines in this region are very faint, and 

 gradually die out in proceeding from the less to the more refrangible 

 side of the spectrum. In the region about L, M, and N the lines are 

 very strong, so that it is for rays of those rates of vibration that 

 nickel is most sensitive at the temperature of the flame. 



Turning now to the base of the flame, we find a great number of 

 lines, of which most extend but a short distance from the bottom of 

 the flame. They form two principal groups, one in the orange and 

 red, and the other in the citron and yellow. These lines are for the 

 most partly sharply defined, and in the more refrangible parts of each 

 group very fine and closely set. They are probably channellings 

 following Rydberg's law, and somewhat confused by overlapping. 

 The diagram indicates the strongest of these lines, as they appear on 

 the background of shaded bands in the flame. It is drawn to a scale 

 of oscillation frequencies. 



None of these lines appear to be nickel lines, and, as they are limited 



