﻿of Light by Metallic Particles. 453 



over phosphorus pentoxide. The pressure was about two 

 inches of mercury. The sparking &c. was carried out as 

 before, and after about four hours' of sparking a fairly good 

 sk} r was visible in spite of the strong fluorescence of the oil. 

 On opening the bottle a decided smell of hydrocyanic acid was 

 noted (nitrogen was known to be present in small quantities), 

 and there were traces of decomposition on the sides of the 

 bottle. An attempt was made to concentrate the particles by 

 standing the bottles on the poles of a Jamin magnet, but no 

 concentration could be observed after twenty-four hours. 

 Considering the viscosity of the oil, this was not regarded as 

 of much weight. The oil was then poured through a filter 

 made from Schleicher and SchulPs special quality paper for 

 stopping fine particles. The paper was washed with kerosene, 

 gasolene, alcohol, and water, and was finally digested in dilute 

 hydrochloric acid. The resultant liquid, on evaporation, 

 showed the presence of traces of iron in an unmistakable 

 manner. The filtered oil was not examined as to its scattering 

 properties, owing to an accident. 



The light scattered from the supposed iron sky behaved in 

 a perfectly normal manner, i. e. like all substances hitherto 

 examined. In this case it was found more convenient to use 

 a beam of sunlight than the light from the arc-lamp ; obser- 

 vations are rendered difficult by the blue fluorescence of the 

 oil. The question turns entirely on the point as to whether 

 the scattering was produced by metallic iron or by its oxide. 

 That the scattering was produced by one or the other I have 

 no doubt ; for after filtering the oil I could discern a faint 

 darkening of the filter-paper which disappeared by treatment 

 with hydrochloric acid, and so disposes of the supposition that 

 the particles were carbon. The iron afterwards found was 

 from the deposit and not originally in the acid or reagents, or 

 in the filter-paper (for check-experiments were made). Some 

 dust on the electrode appeared to be magnetic when tested 

 by a magnetized sewing-needle, but the quantity that I 

 could collect was too small to make the observation conclusive. 

 Of course if I had been in a position to secure a sample of a 

 definite hydrocarbon of small vapour-tension, the matter could 

 easily have been determined ; but such a thing does not exist, 

 to my knowledge, in Australia. 



As I desired further evidence I repeated the whole investi- 

 gation up to the sparking, but this time introduced a small 

 plate of glass between the oil and the electrodes. After 

 sparking as before for three hours (using the coil-contact and 

 three gallon leyden jars with an air-gap) I could see no 

 mirror, though obviously a good deal of oil-vapour had been 



