March i, 1888] 



NATURE 



429 



not only invisible, but are beyond the highest powers of the 

 microscope. It was therefore necessary to adopt soaie method 

 of making them visible. The simplest plan of doing this is to 

 put the air — the particles in which we wish to count — inside a 

 glass receiver, and saturate it with water vapour ; then to super- 

 saturate the air by slightly expanding it by means of an air-pump. 

 When this is done, a fog is produced in the receiver, and we 

 know that each fog particle has a dust particle as a nucleus ; if 

 then we counted these fog particles we would get the number of 

 the dust particles. By this process, however, we would not by 

 any means have counted all the dust particles present, as the 

 fog particles so formed do not represent nearly all the dust 

 particles. If, after time has been given for these fog particles to 

 settle, another supersaturation be made, the receiver will become 

 packed with another set of fog particles, which would require to 

 i>e counted ; and this process would require to be repeated a great 

 number of times before the last particles would become visible 

 and be counted. It is then shown that if there is only a little 

 dust in the air, so that the particles are wide apart, then only one 

 supersaturation is required to make all of them visible. Further, 

 when there are few dust particles present the fog particles are 

 large, and are easily seen falling like fine rain inside the receiver ; 

 and it appeared that if these rain drops could be counted then the 

 solution of the problem promised to be easy. 



The following gives a general idea of the method adopted of 

 working out this suggestion. A small glass receiver w-as con- 

 nected on the one side with an air-pump and on the other with a 

 cotton-wool filter. Inside the receiver was fixed a small stage, 

 about I cm. square, on which the drops were to fall and to be 

 counted. This stage was fixed at a distance of i cm. from the 

 top of the receiver, it was ruled into little squares of i mm., and 

 was examined through the top of the receiver by means of a 

 magnifying glass. To illuminate this stage a gas flame was used, 

 the light being concentrated on it by means of a globular lens 

 full of water. The air in the receiver was pumped out, and 

 filtered air admitted. This air was perfectly dust-free, and 

 gave no condensation when expansion was made. Into this pure 

 air was admitted a small and measured quantity of the air the 

 ]iarticles in which we wished to count. After allowing a short 

 time for the air to get saturated, one stroke of the pump was 

 made, which supersaturated the air, and brought down a shower 

 of fine rain ; while making the stroke with the pump, the stage 

 was carefully observed through the magnifying glass, and the 

 number of drops that fell on a square millimetre counted. This 

 was repeated a number of times, and the average number of 

 drops per square millimetre was obtained, and used for calcu- 

 lating the number of particles in the air. For every drop that 

 fell on the square millimetre, loofell per square centimetre ; and 

 as there is only i cm. of air above the stage that number will 

 represent the number per cubic centimetre in the air of the 

 receiver. Then, knowing the proportion in which the air tested 

 was mixed with pure air, and knowing also the amount to which 

 the air was expanded by the pump, we have all the figures 

 necessary for making the calculation of the number of particles 

 in the airimder examination. 



In constructing the apparatus the first thing to which attention 

 was given was to design the arrangement of stage or j^lalform 

 on which the drops could be most easily seen and counted. The 

 first stage tried was a small piece of glass mirror, ruled on the 

 back into little squares. This seemed at first to give excellent 

 results, the drops being easily seen on its surface ; but on attempt- 

 ing to count them its unsuitableness was at once evident — the 

 confusion produced by the reflected images of the drops caused 

 it to be abandoned at once. Then a mirror of very thick glass 

 was tried, the glass being so thick that the reflected images were 

 out of focus, but it did not give satisfactory results. Very thin 

 mirrors made of microscope glass were then tried, but had to be 

 rejected, because, though they brought the drops and their reflec- 

 tions together, they were unsuitable, being too rough and covered 

 with fine specks on their surface ; only the most highly finished 

 glass can be used for this purpose. The arrangement was then 

 altered, and a transparent stage lit from beneath was tried. This 

 stage was made of a small piece of carefully selected glass, and 

 had the fine lines etched on its surfaces. It was, however, aban- 

 doned, as it did not give such promise as the mirror arrangement. 

 All difficulties in the use ofmirrors were at last got rid of by making 

 them of silver, and now silver mirrors are the only kind used. 

 They are very highly polished, care being taken to keep the 

 rubbing marks in straight lines and in one direction ; they are 

 ruled with fine lines at right angles to each other and at i mm. 



apart. When a silver mirror is mounted in its place, properly 

 adjusted and lighted, it appears, when seen through the lens, 

 like a black surface on which the lines are quite distinct, and on 

 which the small drops shine out brilliantly and are easily 

 counted. 



Some difficulty is experienced in keeping the stage at the 

 proper temperature. If it is too hot, the drops on falling on it 

 do not adhere, but present a beautiful illustration of the spher- 

 oidal condition, as they roll over its surface towards the lower 

 side ot the stage, and drop into the ruled lines, in which they 

 continue rolling till quite evaporated. On the other hand, if the 

 stage is too cold it gets dewed, and counting becomes impossible. 

 Directions are given in the paper for mounting and keeping the 

 counting stage in the best working condition. 



In working the apparatus two methods have been employed 

 of mixing the air to be tested with dust-free air. In one, the 

 dusty air is introduced into a flask which communicates with the 

 test receiver by means of a pipe provided with a stopcock. The 

 small quantity of air that is to be mixed with the pure air in the 

 receiver is displaced from this flask and driven into the receiver 

 by means of a carefully measured quantity of water which is run 

 into the flask. In this way the air to be tested can be measured 

 with a fair degree of accuracy, and as the capacity of the receiver 

 is easily obtained, the experimental errors need not be great. 



In the other method of working, the test receiver is connected 

 with a small gasometer, and the air to be tested is mixed with 

 pure air in the gasometer. The gasometer used has a capacity of 

 20 litres, is carefully graduated and delicately hung, so that the 

 air can be measured in it with a considerable degree of accuracy. 

 In working this arrangement, i litre of the air to be tested is 

 generally first mixed with 19 litres of filtered air. After mixing, 

 nine-tenths of the mixture is let out, and the gasometer again 

 filled up with pure air, and the mixture tested in the receiver. 

 If the drops are still too close, more air is let out, and filtered air 

 added till the desired condition is attained. There must not be too 

 many particles present, or all of them will not fall when ex- 

 pansion is made. Till experience is gained, a check on the 

 quantity is easily obtained by admitting filtered air, in place of 

 the air from the gasometer, and seeing if any drops appear on 

 expansion ; if none, then the correct number has not been 

 exceeded. 



After a satisfactory counting stage had been devised, and the 

 apparatus got into working order, testing began, when at once 

 difficulties presented themselves. The numbers counted in the 

 successive tests of the same air agreed fairly well for a number of 

 times, then all at once the process seemed to break down, and 

 from time to time a great increase in the number was counted, 

 far exceeding the errors of experiment ; then all Mould go right 

 for a time, but only to be followed by failure before long. The 

 first thing suspected for these and for other failures was always 

 the joints of the pipes and the stopcocks, and time after time have 

 the joints been remade with india-rubber solution and stopcocks 

 cleaned and greased, but to find that they are almost never at 

 fault. 



It was then suspected that the failure might be due to the 

 filtered air, with which we mixed the dusty air, not being perfectly 

 freed from its dust. The filtering power of cotton-wool was 

 therefore studied, when it was found that i inch of cotton-wool 

 will filter perfectly if the air is passed very slowly through it, but 

 that even 12 inches of cotton-wool will not check all the particles 

 if the air is made to rush violently through it. Filters must 

 therefore be tested under exactly the conditions in which they are 

 to be used. 



It was, however, found that though the air was only allowed 

 to pass very slowly through even 12 inches of cotton-wool, 

 condensation frequently took place if the expansion and 

 therefore the supersaturation was great. It was thought that in 

 this case the failure might be due to an imperfect action of the 

 filter — that, while it checked most of the dust, yet it allowed 

 the extremely small particles to pass, and that these extremely 

 small particles became active centres of condensation when ex- 

 posed to the high degree of supersaturation used in the tests. Ic 

 therefore here became necessary to test whether the size of the 

 particles has practically any effect on the degree of supersatur- 

 ation necessary to cause the vapour to condense on them. From 

 the investigations of Clerk Maxwell we have theoretical reasons 

 for expecting that the size of the particles will have an influence 

 of this kind, but at present we cannot say that it is sufficiently 

 great to have a perceptible effect in an experiment such as that 

 described. 



