Periodicity of the Spontaneous Ionization of Air, etc. 2l>7 



electrode was connected to one pair of quadrants of! a 

 quadrant electrometer, and also to earth through a vessel, 

 containing air ionized by uranium, which acted as a high 

 resistance. The other pair of quadrants of the electrometer 

 were connected to earth, and in this way the readings of the 

 electrometer were, for small variations, directly proportional 

 to the ionization. During the day the spot of light from the 

 mirror was read on a scale in the ordinary way — the value 

 of the scale readings being found by comparison with a 

 potentiometer. At night the spot was projected on to a 

 clock-driven drum covered with bromide-paper, and the 

 initial and final readings of the potential being known from 

 the scale readings, the intermediate values could be found by 

 measurement of the trace. 



It was thus possible to experiment with any gas in a vessel 

 of any metal, and at first a great many combinations were 

 tried. A zinc vessel was filled successively with hydrogen, 

 coal-gas, carbon dioxide, and air; a lead vessel with carbon 

 dioxide, coal-gas, and air ; and a tin vessel with air. As the 

 apparatus took some time to adjust itself, it was found to be 

 impossible to study the initial changes in the ionization, but 

 a record of the subsequent changes extending over several 

 days, and in some cases over several weeks, was obtained. 

 Curves have been drawn to represent these changes, and it 

 will be seen that they exhibit two main features. In most 

 cases there is a gradual increase of the ionization with the 

 time. This had been previously observed by Burton and 

 McLennan*, but whereas they found the increase to be rapid 

 at first, then to fall away, and finally to cease altogether, 

 leaving the ionization at a steady value, we found a fairly 

 steady increase, which was maintained over a period of ten 

 days or more. It is just possible that this difference is to 

 be explained in some way by the differences in size of the 

 vessels used. 



This increase occurred in the case of lead with each of the 

 gases tried, and so would seem to be characteristic of the 

 metal rather than of the gas. The curves representing the 

 change of ionization with time are shown for lead and carbon 

 dioxide in fig. 1, and for lead and coal-gas in fig. 2. The 

 abscissae are hours of the day, and the ordinates are the cor- 

 responding values of the ionization. Each separate curve 

 represents one day. It will be seen that the curves lie 

 successively above one another and that they differ by ap- 

 proximately equal intervals. In the case of zinc, however, 

 no such increase has been found except in the single case 

 * Phvs. Review, iv. 1903. 



