104 Prof. R. W. Wood on Optical Properties of 



be similar to that of burnished gold, with no trace of 

 diffused white light. It is almost certain that the first bulbs 

 prepared will exhibit colours of inferior brilliancy mixed 

 with diffused white. They frequently improve as fresh 

 deposits are formed by driving the metal back to the clean 

 part of the bulb, and I am not quite certain as to the con- 

 dition most favourable to their formation. The white diffu- 

 sion is due to the presence of a certain number of sodium 

 crystals larger than the ones which form the coloured film. 

 As the thickness of the film increases the colour sequence 

 for interior reflexion is as follows : — straw-yellow, gold, 

 purple, blue. This sequence repeats as the film becomes still 

 thicker, but the colours are now nearly drowned in white 

 diffused light, which cannot be avoided in the case of the 

 thick films. 



If the spectrum of the light reflected from the inner surface 

 be examined, it will be found that only a comparatively 

 narrow region is absent, and that, with increasing thickness, 

 this region moves from the violet towards the red. A series 

 of five photographs of the spectrum of the reflected light is 

 reproduced in Plate I. fig. 4 (Quartz spectrograph and red 

 sensitive plate). This absorption band is very remarkable, 

 and is due, in my opinion, to the circumstance that the spaces 

 or cavities betiveen tlie sodium crystals act as traps for radi- 

 ations of definite frequencies. The colour of the reflected 

 light appears to be nearly independent of the angle of 

 incidence, at least up to very large angles. 



The spectrum of the transmitted light does not show a 

 sharp maximum at the region covered hy the absorption 

 band, though its colour is roughly complementary to that of 

 the reflected light. 



Reflexion from the outer surface of the film, except for 

 very thin films, is non-selective, i.e. white like silver. This 

 circumstance, and the fact that the colour of the transmitted 

 light is independent of the direction of transmission, shows 

 that in the case of incidence on the inner surface, there 

 is selective absorption of a narrow range of the spectrum, 

 which is not present when the light passes in the opposite 

 direction. It is my opinion that the structure of the film is 

 somewhat as shown in fig. 1. 



As I have shown in a previous communication, the conden- 

 sation of a metallic vapour is crystalline, if the temperature 

 of the wall is above a certain critical value, and it is known 

 that sodium crystallizes in octahedra. A structure such as 

 that figured would be expected to show a white reflexion 

 from the outer surface, since the sodium surface is flat and 



