Vapour in the Visible and Ulira-violet Regions. 323 
haye metallic reflecting-power for light of certain wave- 
lengths. For obvious reasons the vapour must be very dense 
if the phenomenon is to be detected experimentally. It must, 
moreover, terminate abruptly in a flat surface. This can 
only be accomplished by confining it in closed vessels made 
of some transparent material, since in a vacuum-tube we have 
a gradual transition from dense to rare vapour at the free 
surface. I feel justified in speaking of the free surface of a 
gas in a vacuum-tube in this particular case. A more pre- 
cise definition of the apparent surface may be an isothermal 
surface, on one side of which we have sodium vapour and on 
the other sodium fog. From suchasurface we should hardly 
expect to get any trace of reflexion, which requires that the 
transition from dense to rare be sudden. If we confine the 
yapour in glass bulbs, we at once encounter many difficulties. 
The glass is at once attacked and discoloured, and we have 
the reflexion from the glass surfaces. An attempt was, how- 
ever, made to observe the phenomenon in the following way. 
A small amount of the metal was sealed up in a hard glass 
bulb, highly exhausted. The bulb was mounted in a small 
air-bath of sheet-iron which could be rapidly raised to a red 
heat. 
The reflexion of the filament of a Nernst lamp by the 
inner and outer surfaces of the bulb was observed through a 
large direct-vision prism. ‘The two minute points of light 
were seen drawn out into spectra, and it was hoped that any 
trace of selective reflexion in the vicinity of the D lines 
would manifest itself by a change in the relative intensities 
of the two spectra at the point in question. 
The bulb was heated very rapidly, but no conclusive ob- 
servation made. The glass discoloured so rapidly that the 
densest vapour could not be studied. In one case it was 
thought that a slight brightening in the yellow occurred, and 
the subject will be further investigated. Possibly by em- 
ploying light polarized in such a plane as to be refused re- 
flexion by glass surfaces, we may be able to get the desired 
results. 
Bulbs of fused quartz become discoloured in the same way, 
but I have been told that fused alumina remains transparent. 
If it be found possible to blow a small bulb of alumina on the 
end of a quartz tube, the question of the selective reflexion 
can be easily settled. | 
The phenomenon is doubtless connected intimately with 
the fluorescence of the vapour. In studying this phenomenon 
with very dense vapours, I was forced to the conclusion that 
the illumination of the vapour with light of the wave-length 
