28 



NUCLEATION OF THE UNCONTAMINATED ATMOSPHERE. 



glass cylinder, while in table 15 there is an irregular distribution of 

 observations between both classes of curves. In table 17 inferior 

 coronas are absent, and those observed present an accentuated case of 

 superior corona. The series fails to detect the large coronas after 

 S/> = 35 cm., so well brought out in table 16. One may note the differ- 

 ent valves used. 



26. Fog limits. An interesting feature of these results are the fog 

 limits or pressure differences at which condensation in dust-free air 

 just commences. In spite of the different sizes of apparatus and 

 valves used, the fog limits are about the same, viz, in tables 13, 14, 

 15, 16, and 17. 



These results are surprising, inasmuch as the effect of the volume 

 ratio of fog and vacuum chambers and the valve effect would naturally 

 be looked to as productive of larger differences. With other appara- 

 tus (Chapter I) the data were : 



Thus the supreme importance of mere rate of exhaustion may well 

 be called in question until more definite results appear ; for with so 

 large a difference of volume ratio, valve obstruction, etc., the essential 

 features should appear more clearly. It is possible that nuclei of 

 extreme fineness (colloidal molecules) may pass through the filter. In 

 such a case these would capture most of the water vapor, reducing the 

 size of coronas by prohibiting condensation on still smaller colloidal 

 molecules. Therefore, if the filter is entirely dispensed with and a 

 closed vessel used, larger coronas may appear at smaller pressure 

 differences. 



