16 J. AITKEN ON THE NUMBER OF DUST PARTICLES IN THE ATMOSPHERE. 



wen 1 due to the imperfect action of the filter. If they were not due to the presence of 

 dust, then we are thrown back on the conclusion that they must have been cases of con- 

 densation without nuclei. In a previous communication* reference has been made to 

 instances of this kind, in which other substances than water are shown to condense with- 

 out nuclei, and experiments are also described in which water vapour seemed to condense 

 without nuclei when the supersaturation was sufficiently great. 



If those troublesome condensations, which took place when high expansions were used, 

 were really cases of condensation without nuclei, then there seemed to be a way of pre- 

 venting them. We know that water may be cooled far below the freezing point and yet 

 it will not become ice, if no nuclei be present ; also that water may be heated far above the 

 boiling point without passing into steam, if no free surface be present. We also know 

 that any violent shock will upset the unstable equilibrium, and determine the formation 

 of ice in the one case, and of steam in the other. Now, it is evident that in these two 

 cases it is more the violence of the motion than the motion itself which determines the 

 change. In both cases there must have been motion in the water due to convection 

 currents while it was cooling or heating, but these slow movements were not sufficient to 

 cause a break in the strained molecular condition. From this it seemed highly probable, 

 that if we checked all violent movements of the air in the receiver, it would be 

 possible to carry the supersaturation to a high degree without spontaneous condensation 

 taking place ; and, on the other hand, we might, by means of violent movements and 

 shocks, cause condensation to take place in supersaturated air though it was free from 

 nuclei. 



Acting on this idea, a number of trials were made with air which had been passed 

 through a long filter, and saturated. On drawing out the piston of the pump with a slow 

 and steady stroke, no condensation ever took place ; whereas a quick stroke generally 

 brought down a shower, and if the piston was made to strike the cover of the cylinder 

 with a sharp blow, the shock invariably brought down a dense shower. These results 

 point clearly to the conclusion that the condensation in the experiments took place without 

 nuclei, as quick and slow strokes have no influence on the condensation when dust is 

 present. It may be mentioned that the expansion was always made quick enough to 

 prevent the air in the receiver absorbing so much heat as practically to interfere with 

 the cooling effect of the expansion. 



Here then was the key to one of our difficulties, accounting for one cause of the 

 variation in the number of particles counted in the successive tests of the same air. 

 Many of the drops counted having no dust nucleus, as it would be inconvenient to 

 regulate the stroke of the pump while attending to other matters, it was necessary to 

 arrange the apparatus so as automatically to check failures of this kind. It has been 

 found that this end can be attained by causing the air on its passage from the receiver 

 to the pump to pass through a small opening, or, what is perhaps better, through a small 

 cotton-wool filter. This filter is made dense enough to check the rush of air, and it also 



* Trans. Roy. Soc. Edin., vol. xxx. part 1. 



