217 
1916-17.] On some Nuclei of Cloudy Condensation. 
amounts of expansion, the bar X carries a slide Y under which there is an 
ordinary boxwood centimetre scale U. On the slide Y is a projecting part 
which stops the pump-handle Z at any desired point. When Z is pushed 
full in and the slide is pushed up to it, the edge of Y is at zero on the 
scale ; so that when the slide is moved to any figure on the scale and the 
expander-handle Z drawn out till it is stopped by it, we know exactly 
what length of stroke we have made, and, knowing the diameter of the 
pump and the capacity of the test-flask and tubes, we can calculate the 
expansion. 
The expander E is connected with the test-flask T by means of the 
tube I. This tube is made up of a short length of glass tube 6 mm. inside 
diameter, and two short lengths of rubber tube which have spiral wires 
inside to prevent them from yielding to outside pressure. The glass part 
of this tube serves the useful purpose of showing if there is any leakage 
in the piston H, the water showing in the tube. If leakage occurs, it can 
be removed by means of the tap R. The substance we wish to test is put 
in the flask V, which is supplied with dustless air by the filter F. When 
testing the effects of heat, the flask Y is removed and a tube put in its 
place between the filter and the test-flask ; the tube is heated by a bunsen 
burner under it, and the substance to be tested put in the tube. 
It may be thought that the tube I connecting the expander E with the 
test-flask is unnecessarily long. There are reasons, however, for the length. 
First, the flask T should be at a convenient height for observation ; 
second, the handle Z of the expander should be at a convenient height 
for quick manipulation ; and, third, the pipes connecting the test-flask must 
have a considerable degree of freedom, as during the tests it requires fre- 
quently to be turned upside down and from side to side, to wet the inside 
walls. If this is not done we cannot be sure the air inside is saturated, 
unless we allow long intervals of time between each expansion. 
As stated, if we know the length of stroke and the diameter of the 
expander pump, also the capacity of the test-flask and its tubes, we can 
