THE STRUOTJTKK OK TITK Nmi.KITS. 73 



nuclei. I am not aware oi" appreciable differences ol' viscosity in the two vapors; 

 but benzine has the smaller latent licut oF evaporation by over seven times. To the 

 extent in which the I'elevant constants of benzol are known, the plieuomena may 

 be provisionally explained with this ratio (^Qf. Oha}). Ill, iji^ 4 and 5). Hence, under 

 identical conditions of nucleation and I'oi- like exhaustions oi' like adiabatic cooling 

 of a given mass of saturated air, the drops would be larger, the colors more 

 advanced in benzine than in water; and since the S(j[uare of radius is in question, 

 this would point to subsidence of the loaded luiclei in benzine nearly four times 

 more I'apid. It would also account for moi'e rapid evaporation or more fleeting 

 colors, which is the case. 



Again, if the loaded nuclei be regarded as mechanical particles, the largest will 

 eventually be found in the lower strata, the smallest in the u})per strata, as in a 

 case of ordinary subsidence of suspended matter in water. It is well known, more- 

 over, that smaller droplets wane, larger droplets grow. Hence, on increasing exhaus- 

 tion condensation takes place first at the bottom and last at the top, since the 

 smallest nuclei correspond to greatest vapor pressure or difficulty in condensation, 

 and since the largest nuclei have been loaded with condensed liquid first, have 

 parted with it last, have had greater time in falling, and have therefore sunk 

 deepest before losing their liquid load. The strata must mount upwai'd as fresh 

 exhaustion proceeds. The last colors to appear are the browns and yellows of the 

 first order, like those seen in the steam tube for vanishing condensations. The 

 whole phenomenon is thus the result of strata of invisible nuclei, (jradedin virtue 

 of the loading mechanism, and partakes throughout of a mechanical character to the 

 extent that the nuclei are not even a uniform product. The forced disti-ibutiou is 

 sufficiently powerful to entirely mask the elementaiy optical phenomenon. 



On shaking the liquid benzine in the receiver, uniform distribution is again 

 promoted, with the result that annular coronas reappear. It is particularly to be 

 noticed that subsidence is due to loaded nuclei. The free nucleus does not 

 appreciably descend. Even with water vapor, loading does not produce stratifica- 

 tion. Water fogs when exceptionally dense may sometimes be seen to rise, but the 

 diffraction pattern is always annular and usually without color distortion. 



6. After completing the above woi-k, I made similar experiments with benzol, 

 reaching the additional result that the nuclei hei'e behave as if they were produced 

 by the liquid itself in the dark. The sequel will show that they are not really so 

 produced, but have a habit of brooding over the surface and thus escaping detection 

 until revealed by diffusion. They then ascend against gravity in hoi'izontal strata. 

 They may be precipitated by partial exhaustion, leaving the air neai'ly free from 

 nuclei, but the above flask was refilled with nuclei in 10 or 20 minutes. The 

 experiment may be repeated any number of times. The shai-p demarcation 

 between the pure air above and the rising surface of nuclei below, is beautifully 

 evidenced by the coronas, which are annularly perfect for axial beams below the 

 surface, asymptotically bowl-shaped near the surface, and absent for axial beams 

 above the surface. Cf. figure 2, Chapter 111. 



Truly spontaneous nucleation in the dark was only discovered in a few cases, 



