157 
1888 - 89 .] Mr John Aitken on Dust Particles. 
At first it was difficult to say whether the difference in the numbers 
counted was due to a real difference in the impurities of the air. A 
greenhouse fire, at a considerable distance to the windward, suggested 
it might be so, and on some occasions, no doubt, there was a real 
difference due to this cause ; but as the great variation continued 
after the wind changed, another explanation had to be found. It 
was then noticed that the manner of admitting the air to the receiver 
had an influence. After repeated trials, it was found that if the air 
was admitted with a rush far fewer drops were counted than when 
it was allowed to flow in slowly ; but which, if either, of the numbers 
counted under these two conditions was the correct one'? It had 
previously been observed that if the air was allowed to rush in, 
and in its passage strike the surface of water, nuclei w'ere manu- 
factured, and generally that violent currents of air have a tendency 
to produce centres of condensation. But in the case above referred to, 
the slowest moving air gave the greatest number of drops. Were 
nuclei formed under some unknown conditions ? or was it possible 
that when the air enters with a rush the expansion in the entrance 
tube is sufficient to cause condensation before entering the receiver, 
if the air is moist? To settle this point, the method of working was 
altered. The air was not allowed to go into the receiver immediately 
after expansion was made, but water was allowed to enter by the pipe 
T. The stopcock by which the air enters was then opened full, and 
the outflowing water drew the air slowly into the receiver, through 
wide open passages and without a rush. By this plan we could trust 
that no particles were formed by the movement of the air, and no 
expansion produced in the entrance pipe sufficient to cause condensa- 
tion in even nearly saturated air. When tried, it was found that the 
numbers obtained by this plan of working corresponded with those 
got when the air was admitted slowly. The numbers, therefore, 
obtained when the air was allowed to rush in were too small, and 
the smallness must have been due to the air losing some of its dust 
in the process by forming raindrops while still in the entrance pipe. 
The time required to make a test need not be great ; indeed, con- 
sidering the numbers we have to deal with, we may say it is very 
short. To put the apparatus together and ready for work need not 
take more than five minutes, more or less, according to the expertness 
of the operator. After it is in working order, half a minute is quite 
