156 
Proceedings of Royal Society of Edinburgh. [sess. 
then we had better restrict our attention to 4 squares. For general 
work, however, 5 per square mm. is the top, and 1 per square mm. 
the lower limit. It is for this reason that in the set of stopcock 
measures described, their capacities have been fixed at something like 
this proportion. The sizes of these measures are 5 c.c., 1 c.c., \ c.c., 
and c.c. The second last one is not 1- of the next largest, but it 
is of a size that makes the calculations easy. The largest measure 
requires the number per c.c. counted on the stage to be multiplied 
by 100, the next by 500, the next by 2000, and the last by 10,000. 
It may be as well here to call attention to the necessity of making 
all joints in the apparatus perfectly air-tight, and all stopcocks must 
be of first-class workmanship, as it is found that any leak that allows 
air to pass, also allows dust to get in along with it. No hole between 
metal surfaces seems to be too small for dust to pass through, or is 
it that no hole that cannot pass dust will pass air, owing to the dust 
particles sticking and closing the passage to the air % 
I shall now refer to a precaution necessary in working this 
apparatus. If the air in the place where we are working is damp 
and nearly saturated, care must be taken in opening the receiver 
stopcock. It must be done very gradually, so that the air may enter 
slowly. If opened to the full extent suddenly, the dusty air in 
the entering pipe is expanded, condensation takes place on the 
dust, the drops formed in the pipe are driven violently by the 
rush of air, and many adhere to the surfaces with which they come 
in contact. This effect only takes place when the air is very damp, 
otherwise the expanded air is not cooled to its condensing point 
before entering the receiver. This condensation in the entering pipe 
is greatly due to the resistance of the cotton-wool filter preventing the 
air rushing in to supply the partial vacuum. A filter of large area 
is therefore desirable as a prevention of errors of this kind ; but as 
even a large filter will not entirely prevent condensation, a little care 
must always be taken to open the receiver stopcock slowly when the 
air is damp. 
The difficulty above referred to was met on the first occasion on 
which I made tests away from the laboratory. These tests were 
made in winter in an outhouse, and when the atmosphere was very 
damp. In a great number of tests the numbers counted on the 
stage varied to an amount far exceeding any previous experience. 
