Resonance of Metal Paii ides for Light-Waves. 259 



a further assumption, and that might take the form of 

 assuming that the number of systems which in time "dt pass 

 out of the extension in phase da and into da' is proportional 

 to the number that are in da for the time being. Other 

 assumptions might answer the purpose. But as Professor 

 Gibbs has left it, 77 seems, according to the definition you 

 2'ive, either constant during all periods of time however long, 

 or variable during any period of time however short. 



[The above article was sent to press before the author was 

 aware of the lamented death of Professor Willard Gibbs.] 



XXXI. The Electrical Resonance of Metal Particles for Light- 

 Waves. Third Communication. By R. W. WooD^ Pro- 

 fessor of Eicperimental Physics^ Johns Hopkins University *. 



SINCE the appearance of the two preceding papers (Phil. 

 Mag. vols. iii. & iv., 1902) on the above subject I have 

 made further investigations, which appear to confirm the pro- 

 visional hypothesis of electrical resonance, which was adopted 

 to explain the very brilliant colours of films made up of metal 

 granules of the order of magnitude of light- waves. 



I have succeeded in obtaining the coloured films in pris- 

 matic form, and have established the fact that they exhibit 

 anomalous dispersion for waves longer and shorter than the 

 ones which are refused transmission. This was observed for 

 electrical waves passing through a prism built up of tinfoil 

 resonators by Garbasso and Aschkinass. I have obtained 

 the coloured films on the walls of a tube in which I had fused 

 a quantity of magnalium alloy in vacuo. The film was per- 

 manent in air, but the colours vanished as soon as the film 

 was moistened with a little very dilute acid, owing to the 

 conversion of the metal granules into a salt. I have also 

 obtained coloured films by the cathode discharge from an 

 electrode of selenium, though the colours in this case were 

 not nearly as brilliant as those obtained by the distillation of 

 the alkali metals. The microscope showed these films to be 

 granular, and experiments with the quartz spectrograph 

 demonstrated that they were very transparent to ultra-violet 

 light, for which homogeneous films of selenium of the same 

 thickness are absolutely opaque. 



The silver films, red, purple, and blue, which I described 

 in my second paper, seemed most suitable for continuing the 

 experiments, owing to the comparative ease with which they 



* Communicated by the Physical Society. 



S2 



1 



i* 



