Dowi.iNG & Haughev — Electrification, Phosphorus Smoke Nuclei. 59 



XVII. — We now turn to the consideration of the observations carried 

 out under the lowest fields, as described in section 10. It will be noticed 

 that the first "step" is completely missing, and that the second is con- 

 siderably modified. There is not any good reason why the simple "whole 

 n\imber " progression should not hold right down to the commencement, so 

 it was necessary to find some explanation for this obliteration of the pre- 

 liminary steps. It is natural to expect that the ionisation effect should be 

 important under low fields. Although the apparatus (fig. 6) was designed 

 with a view to the ions being removed by the field before the smoke stream- 

 yielded up its charge, nevertheless under the weaker fields this process is 

 probably incomplete. To ascertain how this " ion " current would affect our 

 results, we covered the phosphorus pellet with a bridge of metal foil, so that 

 the air could pass across as before, while the phosphorus, was practically 

 shielded from the electric field. Fig. 8 shows the form of curve obtained 

 I'to a different scale to fig. 7). Little, if any, charge can be now carried by the 

 smoke, and the curve indicates that the ionisation current reaches a satura- 

 tion value at about six volts. The magnitude of the ion-current is then 

 probably greater than that due to the singly charged smoke, and the bend at 

 saturation would occur about the middle part of the first " step." The result 

 would clearly be to flatten out the curve (fig. 7) along the first and part of the 

 second step, which is exactly what is usually found. 



SinoiAEY AND Conclusions. 



XVIII. — It has been shown that, when smouldering phosphorus is 

 exposed to an electric field, the smoke particles take up electric charges, of 

 which the magnitude is determined by the field intensity. Eeasons are given 

 for the assumption that the particles approximate to a uniform size. 



The periodic increase of the charge, due to successive electrons, is dis- 

 played in the observations, and enables us to determine the number of 

 electrons assumed by each smoke particle irnder given conditions. 



All values of this number have been observed from unity to twenty-five, 

 and iro doubt the range could be extended. 



If an accurate " count " of the smoke particles could be made, it would be 

 possible to base a determination of f on the principles of this work. 



An explanation is put forward for the apparent stability and eohesiuu uf 

 the smoke filament, and an apparent discrepancy between this woik and 

 Millikan's is accounted for. 



R.I.A. PROC, VOL. XXSTI, SKCT. A. [7] 



