A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 53 



If the exhaustion chosen is such as to reduce the vapor pressure to \ 

 (this corresponds roughly to a pressure difference of ^p'^iy), the intersection 

 of the vertical line of the figure with the successi\'e curves will show the heights 

 of the fog-banks on condensation. Thus after 20, 40, or 60 minutes the fog- 

 baixks, having attained heights of, roughly, 22, ^t„ and 40 centimeters. wiU be 

 in good position for observation. 



For any other vapor than water, the times will increase inversely as the 

 coefficients of diffusion. Thus for benzol the time intervals should be increased 

 about 2 1 times. 



4. Conclusions. — The striking feature of these curves is the extreme slow- 

 ness of diffusion even for water vapor. At but 20 centimeters above the liquid 

 surface it takes half an hour to reach semi-saturation. The case is accentuated 

 for other liquids where the coefficients are smaller, as, for instance, for the follow- 

 ing liquids at about 20°: 



If, thei-efore, the fog particles are relatively numerous, large, and subside 

 rapidly, the air will soon become highly de-saturated. In other words, if the 

 air in the receiver A is cleaned of nuclei by condensation, there is no vapor 

 available to replace the moisture lost. 



In case of water vapor the fog particles are small and subside slowly while 

 the vapor is lighter than air. Hence the latter is liable to be reheated from 

 the rapid radiation of gases assisted by convection, as stated, before much 

 de-saturation takes place, unless the vessel is very long and the sides dry. Pre- 

 cisely the opposite is the case for the hydrocarbon vapors in spite of their vola- 

 tility, since the fog particles for the same nucleation are larger and fall rapidly, 

 and where the vapors are heavier than air. After successive precipitations at a 

 given pressure difference the vapor may be so far de-saturated that it nearly 

 ceases to condense even if nuclei are present. That it can quite cease to re- 

 spond is impossible, for some vapor must return to the air after condensation 

 almost instantaneously; but it is not improbable that a vapor exhausted to a 

 slightly higher pressure difference will fail to respond thereafter at the original 

 presstire difference. Thus there is considerable chance for error, and what is 

 taken for the diffusion of nuclei added near the surface of the liquid may actu- 

 ally be the dift\ision of the liquid itself. This will even be the case if an aspira- 

 tion cun-ent, as in § i, falls sufficiently short of saturation, supposing always 

 that the velocity of nuclei is relatively large. True, in the experiir.ents which I 

 made, the two sets of results for nuclei and for vapors differ radically in order 

 of values, in distribution among different vapors, while for carbon disulphide 

 the gradually increasing aperttires of the coronas is certain evidence of greater 

 concentration of nuclei. But these and other occurrences may each in their 

 turn be explained away. 



5. Diffusion from greater to less saturation. — ^To facilitate the discrimination 



