4 2o SCIENCE PROGRESS. 



minated region, but there appeared nothing metallic about 

 it, and it was easily dusted off. It seemed to be merely 

 dust out of the air. So the experiment was repeated in a 

 dust-free chamber, containing air filtered slowly through 

 Ions' tubes of cotton-wool, and now not the faintest local 

 dimming of the surface could be observed, although the illu- 

 mination and electrification lasted for days. So the answer 

 for silver was in the negative. 



Next a non-magnetic substance was hung in a powerful 

 converging magnetic field in the neighbourhood of clean 

 illuminated and oppositely electrified iron, to see if by con- 

 densation of evaporated iron, it was possible that it became 

 magnetic. A minute torsion-bar of copper suspended over 

 a clean, conical, vertically pointing electro-magnet's pole 

 was the best arrangement. There were difficulties about 

 this experiment on account of electrodal and other forces, 

 but so far as disturbance could be eliminated the result 

 for iron was also negative. 



Then the most elaborate series of observation was made 

 on metallic sodium kept in an atmosphere of highly purified 

 hydrogen, the gas being supplied through a long series of 

 drying tubes, and kept burning as a small jet just after it 

 had passed over the sodium surface. By a mechanical 

 arrangement the sodium could be cut to a clean surface from 

 outside, and when the gas was pure this surface lasted a 

 fairly long time, and under illumination it discharged elec- 

 tricity supplied by several dry piles in series, so that a 

 considerable supply of electricity could be drawn from the 

 flame whenever light from an arc lamp was allowed to 

 fall on the sodium surface through a quartz window. The 

 flame was looked at either direct or through a small 

 spectroscope, and though the sodium line could not be kept 

 wholly absent, its occasional presence depended in no way 

 on whether the surface was positively or negatively or not 

 at all electrified, nor on whether the light was or was not 

 shining on it. 



Hence I conclude that the discharge of electricity from 

 illuminated surfaces is not effected by evaporation of those 

 surfaces, but that the molecules which convey the charge 



