﻿708 Prof. R. W. Wood on Selective Reflexion, Scattering 



is much to be desired, and could be easily tested by experi- 

 ment. The calculation could be made for a beam of either 

 rectangular or circular cross-section. I believe that a beam 

 with a square cross-section will prove best, and it must be 

 remembered that in the experiment we photograph the entire 

 mass of the gas, as seen from the side. I found that with a 

 beam of this description the intensity of the secondary radia- 

 tion close to the beam of primary radiation was in some cases 

 equal to one quarter of the value of the latter, though it fell 

 off very rapidly in value as the distance from the primary 

 beam was increased. 



It thus appears that the destruction of the secondary 

 resonance radiation is the result of the introduction of the 

 factor of absorption, but it seemed best to test other hypo- 

 theses if possible. The only alternative that I have been 

 able to think of is the possibility that the presence of the 

 air alters the wave-length of the emitted light slightly so 

 that it is no longer capable of exciting the vapour. The 

 change necessary to accomplish this would be very slight, too 

 small in all probability to be detected by any spectroscope 

 except those of the highest resolving power. Another method, 

 equally satisfactory and involving less difficulty than the use 

 of a spectroscope, occurred to me. This we may call the 

 method of the differential vacuum. A double cell with a 

 quartz partition was constructed, of the form shown in 

 PL X. fig. 6. Each compartment contained a drop of 

 mercury, and the monochromatic light from the arc was 

 focussed at the centre of the lower compartment, which was 

 exhausted to a pressure of 6 mm., previous experiments having 

 shown that the secondary radiation is practically unnoticeable 

 at this pressure. The upper compartment was highly ex- 

 hausted. The point at which the rays were focussed was 

 brought as close as possible to the under side of the hori- 

 zontal partition separating the two cells, and as close as 

 possible to the window through which the radiation was 

 photographed, to avoid the loss of light due to an intervening 

 layer of vapour. If now there is a change of wave-length 

 due to the air in the lower compartment, the light (emitted 

 by the directly excited vapour) which passes up through the 

 quartz partition will be unable to stimulate the vapour in the 

 upper compartment. If, on the contrary, the absorption 

 hypothesis is correct, the vapour in the upper compartment 

 will glow, since we have eliminated absorption here by the 

 complete removal of the air. The photograph showed the 

 secondary radiation in the upper compartment, and the ratio 

 of its intensity to that of the primary was about the same as 



