150 NUCLEATION OF THE UNCONTAMINATED ATMOSPHERE. 



proportion as the molecular dimensions are approached. But even 

 when the yellows of the first order vanish, condensation probably still 

 takes place on the colloidal molecules specified. It is natural to asso- 

 ciate these extremely fine nuclei with the existence of a very pene- 

 trating radiation, known to be present everywhere. Moreover, the 

 occurrence of many nuclei with but few ions is not contradictory, if 

 the latter are only manifest when the former are made or broken, in 

 the manner suggested above (section 89). 



111. The nucleation Of filtered air. If the filtration is moderately 

 slow, and if the pressure difference, Bp, continually increases, the 

 angular coronal diameter or its equivalent, s, terminates in a horizon- 

 tal asymptote, as shown, for instance, in Chapter III, figure 45 et seq. 

 Hence the number of efficient nuclei in the exhausted receiver eventu- 

 ally approaches a constant, specific for the given rate of filtration. 

 It is probable, therefore, that extremely small nuclei or colloidal mole- 

 cules (very small even when compared with ions) pass through the 

 filter ; for in such a case more nuclei would enter the fog chamber 

 during the influx of filtered air to replace that removed by exhaustion, 

 in proportion as this exhaustion (8/>) is higher. Hence s should be 

 constant in the manner actually observed. The study of the succes- 

 sive groups of nuclei in a scale of decreasing smallness promises to be 

 interesting. 



If the current of air through the filter is successively decreased until 

 its velocity all but vanishes, the asymptote in question may be raised 

 enormously until the curve runs upward with a nearly straight sweep. 

 Thus, for extremely slow influx of filtered air to restore the normal 

 pressure after exhaustion, values like the following appear : 



5^ 24 30 33 37 41 



s = rain 5.4 5.8 7.2 7.3 

 N X io~ 3 = i 105 143 280 351 



data which, from the nature of the work, are inevitably somewhat irreg- 

 ular, but which do not even suggest an asymptote, and from which 

 the character of figure 46 has departed. These nuclei can not come 

 through the filter, for which case ^ = coust. would be conditional. The 

 fact may also be proved by making observation but once in 24 or 48 

 hours, in which interval all nuclei originally present would vanish by 

 time loss, unless constantly reestablished as a case of molecular equi- 

 librium, as already suggested. It is possible to filter slowly enough 

 that, ccet. par., a specific nucleation may appear for each pressure 

 difference. 



Experiments have been in progress in this laboratory since May 9, 

 in which the nucleatiou of filtered air is examined daily with regard to 



