J. AITKEN ON THE NUMBER OF DUST PARTICLES IN THE ATMOSPHERE. 9 



The outside air is drawn through the pipe for some time just before it is connected with 

 the gasometer. One litre of air is drawn into the gasometer which had been previously 

 nearly filled with filtered air, and filtered air is then added to complete the 20 litres, the 

 whole is stirred, after which the stopcock attached to the gasometer is opened, and ^ of 

 the air allowed to escape, after which the gasometer is again filled up with filtered air. The 

 gasometer is now connected with the test receiver by means of the tube E, the air in the 

 test receiver having been previously purified by means of a filter. The air from the 

 gasometer is now made to flow through the test receiver for some time to expel the 

 pure air, the stopcock is then closed, expansion made, and the drops counted. Or, 

 instead of passing a great quantity of air through the receiver so as to expel all pure air 

 from it, we may allow only just as much to enter as will supply the place of what was 

 taken out by the one stroke of the pump when the expansion was made. These two ways 

 of working, of course, give different numbers of drops per millimetre, but, when the 

 calculations are made for the different conditions, the numbers are alike. 



The dusty air in the gasometer was, by the process described, first mixed with 19 times 

 its volume of pure air, and then T \y was mixed with 9 times its volume ; so that the 

 original air was diluted 200 times. This proportion is found to suit some states of the 

 outside air ; but for air with more particles in it, such as the air of a room, the process 

 has to be repeated twice or even three times, each process of emptying to ^, and filling 

 with filtered air, requiring the number of drops to be multiplied by 10. The gasometer, 

 however, is not always emptied to -j\j ; the density of the shower which appears when a 

 trial expansion is made gives an idea of the amount of dilution required before correct 

 counting can be done. It would save time with very dusty air to let out more than T 9 a 

 before filling up with pure air, but it is not thought advisable to do so, as the measure- 

 ments made with the lower part of the gasometer are never so accurate as those taken 

 further up, owing to evaporation altering the level of the water. In some cases the air 

 to be tested has been introduced by displacement, the air being contained in a large 

 bottle and displaced by means of a measured quantity of water, attention being given to 

 keep pressure correct. By this method of working, we can mix very small quantities of 

 the dusty air, and make the test without emptying and filling the gasometer a number 

 of times. 



On some Necessary Precautions. 



In developing a process of this kind, it was quite to be expected that many difficulties 

 would present themselves. Most of these have now been overcome, but more than once 

 the task appeared hopeless. Everything would go right for a time, then all at once it 

 would break down. The number of particles counted in the successive tests of the same 

 air would be fairly constant for a number of times ; it would then increase to an amount 

 far exceeding the limits of the errors of observation. Now that the road is cleared and 

 the principal obstructions removed, it is difficult on looking back to see where the 

 obstructions were, as their cause, and the manner of avoiding them, are now so 



VOL. XXXV. PART 1. B 



