OONDENSATIOTSr NUCLEI. 199 



LIINIITING SITPERSATURATIOX IN DITST-KKEE (iASES. 



When air saturated "with watei- vapor lias been freed from (hist 

 particles no drops are formed on expansion, provided that a certain 

 critical degree of supersaturation has not been exceeded. To pro- 

 duce the supersaturation necessary for condensation in the form of 

 drops in dust-fr(H' air the air must be allowed to expand suddenly, 

 till the final volume is 1.25 times the initial volume. The condensa- 

 tion is rainlike in form, and the number of dro|)s remains small 

 although the expansion considerably exceeds this lower limit. Ex- 

 pansions exceeding a second limit, /'y('\=l.;^S give fogs, which increase 

 rapidly in density, i. e., in the mnnber of the dro[)s as the expansion 

 is increased beyond this limit. In such experiments it is of course 

 necessary that the apparatus used should be such that a very rapid 

 change of volume can be brought about, and that the ratio of the final 

 to the initial volume is known with certainty. Some years ago I 

 introduced a method which has proved suitable for the purpose. 

 When this method is used it would appear from the consistency of the 

 results obtained with cloud chambers varying in capacity from 15 to 

 1,500 cc. that the expansion is adiabatic and is completed before any 

 appreciable quantity of water has had time to separate out. From 

 the ratio of the final to the initial volume, knowing the initial temper- 

 ature, we can deduce the temperature at the moment when the expan- 

 : ion was completed from the equation for the cooling of a gas by 

 adiabatic expansion. 



5, \vj 



Y-\ 



y may be taken as not ditfering sensibly from its value for the dry 

 gas. Knowing the final temperature we have the data from which we 

 can obtain the density of the vapor which would be re(iuired for sat- 

 uration at the moment of completion of the expansion, and we know 

 the actual density at that moment from the initial temperature and 

 the ratio of the final to the initial volume. Thus the supersaturation 

 measured by the ratio of the actual density of the vapor at the instant 

 wdien the expansion is completed to the density of the saturated vapor 

 at the temperature which thi' su])ersaturated gas (hen possesses can 

 be calculated. 



The supersaturation required for the rainlike cond'Misation is found 

 in this way to be approximately fourfold, that recpiired for the cloud- 

 like condensation being nearly eightfold. There are these two classes 

 of nuclei always present in moist dust-free air, and always being pro- 

 duced, for however often the process of condensing water on the 

 nuclei and allowing the drops to settle is repeated the mnnber of 

 drops formed in subsecpient expansions is undiminished. The nuclei 



