THE STIU'CTIUM.; (11 TIIK .NUCLPilJS. 139 



lower li;ilf ul" the receiver full of nuclei, or to bring .-ihout other similarly stratified 

 conditions evidenced by semi-coronas, , quarter coronas, etc. Indeed, when foreign 

 nuclei are introduced they usually appear, on exhaustion, to be distril)uted In 

 layers, usually in couches immediately over the sui'face of the liquid. Ciiai)ter IV, 

 §§ 17-20, 23-26. The rates of difl'usion are here easily measured, whereas in case 

 of the electrolytic solvents like watei', measurement is difficult, not only because 

 diffusion is over 100 times faster, but l)ecause the diffusion of nuclei into pure air 

 is a l)ranched design, like the roots of a tree. Finally, unless the vapor tension of 

 the licxuid is too small (like petroleum), only the normal coronas of Chapter II, 

 § 6, need be expected to occur in the cases of the vapors of non-electi-olytes. For 

 it is not unreasonable to associate large coi'onal particles with the coi'respondinf 

 large nuclei. 



3. Effect of the medium. — It follows from this that the nuclei derived from 

 the same source, and undei- otherwise like conditions, must be relatively small in 

 case of a vapor of the fii'st group of li(piids, and relatively large in case of a vapor 

 of the other liquids. The same nucleus, so far as its apparent origin is con- 

 cerned, differs in size with the medium in which it is suspended or in which it is 

 generated. 



[On reading the proofs of this chaptei-, I observe that a discussion of the dif- 

 fusion ujnvard of the vapor itself into air has not been inserted. This dift'usiou is 

 so much slower than would correspond to the data observed for nuclei — it cori'e- 

 sponds to a quadratic law, while the order of vapor diffusion is not identical with 

 the ordei's of h found in this chapter — that I ovei'looked it. The dift"nsion line for 

 benzol, for instance (flg. 4, below), is about 40 times in excess of the values which 

 would correspond to the upward motion of an 85 [)er cent vajmr saturation, into 

 air. Again, in the case of carbon disulphide, Chapter IV, § 26, the dift'usion of 

 nuclei out of the liquid is tested in terms of the apei'tures of the coronas. These 

 increase in the coui-se of time, showing a gradual increase of nucleation for the 

 same vapoi- pressure. An increase of vapoi' pressure for the same nucleation would 

 pi'oduce larger fog particles and smaller coronas in the lap.se of time, the reverse of 

 what was obsei'ved. 



Nevertheless, from computations which I have since made at some length, it 

 is necessary to admit, that if what has been interpi'eted as the diffusion of luiclei 

 at approximately constant vapor pressure, is the dift'usion of a definitely under- 

 saturated vapor, at approximately constant nucleation, large diffusion i-ates are not 

 excluded, and the inferences as to luiclear velocity have been ro})bed of their force. 

 Thus, for I)enzol, the observed data would correspond to the dift'usion of a vapor 

 at about 1/3 saturation, into homogeneously nucleated air. In fact, by varying the 

 pressure dift'ereuces on exhaustion, one should be able to test the progi-ess of the 

 diffusion of a vapor into nucleated air throughout the whole extent of the column. 

 Hence the argument of the present chapter must be read with this alternative 

 explanation in view, and I shall have to return to the subject moi-e sj)ecifically 

 elsewhere. In all cases, however, the velocity of the nucleus nuist be in excess of 

 the velocity of the obsei'ved plane of vapor.] 



