206 
Proceedings of the Royal Society of Edinburgh. [Sess. 
the instruments. The passage of a train is marked, and the effect of its 
smoke came several minutes later. 
Train 1. No visible smoke. 
„ 2. A little white smoke. 
„ 3. Copious white smoke. 
„ 4. No visible smoke. 
„ 5. No visible smoke. 
„ 6. No visible smoke. 
7. Two trains 
(a) Dense black smoke. 
( b ) No visible smoke. 
No effect was recorded, probably owing to some slight variation in the 
wind. 
Train 8. White smoke. 
„ 9. White smoke. 
10. Two trains 
(a) Dense black smoke. 
( b ) No visible smoke. 
In the case of the white smoke the colour is due to moisture, and all 
whiteness had generally disappeared long before the smoke reached the 
instruments. Wilson concluded that the positive ions in a bunsen flame 
consist of charged molecules of the gases present. Similarly solid particles 
do not seem to be necessary for the carriage of the positive charge in 
smoke. 
These potential gradient measurements confirm the conclusions of others 
that by combustion a considerable amount of ionisation is produced ; but 
as the effect is always to produce an increase of the positive potential 
gradient, more positive than negative ions may be formed. 
By burning considerable quantities of benzene, methylated spirits, and 
sulphur, separately and simultaneously in the open, at distances up to 
25 yards from the collector, and under various meteorological conditions we 
were unable to reproduce the smoke effect. On burning these substances 
in the laboratory we found that the cooled combustion gases, in each case, 
contained both positive and negative ions,* and in approximately equal 
numbers as far as we could gauge with the apparatus at our disposal. 
The mixed products of combustion would contain C0 2 , S0 2 , S0 3 , carbon 
particles, water vapour, and nuclei both of condensation and of charge, 
as in the case of coal smoke. 
We consider, therefore, that the ionisation giving rise to the largely 
* L. Bloeh, Annates de Ghemie et de Physique, xxii. and xxiii. Reoglie and Brizard, 
Comptes Rendus, 1909, p. 146. 
