rrox SOME oe the phenomena oe combustion. 
637 
enabled me to ascertain, by photometrical measurements, the extent of this loss 
of luminosity in rarefied ah’. The result proved that a great reduction in illumi- 
nating effect ensues when a candle is transferred from air at the ordinary atmospheric 
pressure to rarefied air. At the same time remarkable changes in appearance occur in 
the fiame itself, especially at high degrees of rarefaction. During the diminution of 
pressure down to half an atmosphere, the chief alteration is the gradual invasion of the 
upper and luminous portion of the fiame, by the lower blue and non-luminous part. As 
the pressure sinks towards 10 inches of mercury, the retreat of the luminous portion of 
the. flame towards the apex goes on uninterruptedly, but the shape and colour of the 
flame also begin to undergo very remarkable alterations ; the summit becomes more and 
more rounded, until at 10 inches pressure the flame assumes nearly the form of an ellipse, 
whilst the blue portion, which now comprises nearly the whole flame, acquires a peculiar 
greenish tint. Finally, at 6 inches pressure the last trace of yellow disappears from the 
summit of the flame, lea-vlng the latter an almost perfect globe of the peculiar greenish- 
blue tint above mentioned. Just before the disappearance of the yellow portion of the 
flame, there comes into \lew a splendid halo of pinkish light, forming a shell half an 
inch thick around the blue-green nucleus, and thus greatly enlarging the dimensions 
of the flame. The colour of this luminous shell closely resembles that first noticed 
by Gassiot in the stratified electrical discharge passing through a nearly vacuous tube 
containing a minute trace of nitrogen. The colour thus imparted to the electrical 
discharge undoubtedly constitutes the most delicate test of the presence of nitrogen. 
In both cases I believe the coloured light to be due to incandescent nitrogen. Under 
a pressure of 4 ’6 inches of mercury, a small gas-flame burning in the chimmey 
Plate XVIII. fig. 1, nearly fills the latter with the pinkish glow just mentioned, which 
extends to a height of nearly 3 mches above the true flame, forcibly reminding the 
observer of the electrical discharge through a nearly vacuous tube. The gas-flame did 
not manifest any tendency to extinction at this low pressure. 
In attempting photometrical determinations with candles, it was found that, owing to 
the irregularities of combustion already noticed, no satisfactory quantitative experiments 
could be made in artificially rarefied air. Oil-lamps proved also equally unsuitable, owing 
to the gradual ascent of the base of the flame towards the apex of the wick, by which 
the size of the flame and the hourly consumption of oil were greatly diminished. Ke- 
course was therefore had to coal-gas, which, although liable to certain minor disturbing 
influences, yet yielded results, during an extensive series of experiments, exhibiting suf- 
ficient uniformity to render them worthy of confidence. 
Fig. 1, Plate XVIII., represents the arrangement of the apparatus employed. A is a 
governor into which the gas was first led, and whence it issued through a T-piece, one 
branch of which led to the jet B, which I will call the standard flame, whilst the other 
communicated with the test-meter C supplying a jet D, which may be conveniently termed 
the experimental flame : thus the delivery of gas of uniform pressure, at the two stopcocks 
regulating the supply to the two flames, was secured. The standard flame was shielded 
4 R 2 
