72 THE STRUCTURE OK THE NUCLEUS. 



they VHuisbed from sight, hiouii, yellow white heing the last colors observed. 

 Uniform color fields (strata of limiting width) were eventually producible in this 

 way. Yellow, brown, ciirason, ai'ose from a whitish blue base, then descended 

 again on completed exhaustion, reminding one of the extension of an accoidion. 

 The speed of apparent viscous subsidence of the top bands has no direct meaning, 

 since fall (or J'ise) is here complicated by evaporation. 



On entrance of air, voi-tices were evidenced by ring-shaped threa<ls of color so 

 that mixture wa.s at first in('vitaV)le. One must wait till this ceases before again 

 exhausting. Convection currents due to local reheating of the adiabatically cooled 

 gas by the walls of the receivei', were equally apparent, stringy colors lising on the 

 outside and descending into the middle of the receiver. It is the same phenomenon 

 which interferes with the usefulness of narrow tubular appai'atus. 



4. As this subsidence of color bands in benzine vapor is an observation of 

 importance, I resolved to repeat the work under more normal conditions. Accord- 

 ingly, I used as my source of light the bi-ight area of the mantle of a Wels])acli 

 burner, seen through a small hole in the metallic screen by an eye, looking centrally 

 thi'ough the receiver containing satui'ated benzine vapor and nucleated air. Punk 

 nuclei re[)laced the phosphorus nuclei. On exhaustion (without nucleationj after 

 standing overnight, the coronas were white centered, fringed with blown, about as 

 large .as ordinary lycopodium coronas seen under like conditions. These large 

 drops are a proof of the relative absence of nuclei initially. 



After nucleation, the first dense fogs were vaguely annular during the first five 

 successive exhaustions, filtered air being supplied between each. The next five 

 exhaustions {)ioduced more nearly, finally very fully, stratified colors, in spite of the 

 point source of light. Shaking the receiver violently at any time, so as to scatter 

 the liquid benzine within, always reproduced a nearly perfect corona, which, on 

 standing, became distorted again, in color at least. 1 now made special experiments, 

 shaking the receiver before each observation, bringing out successive coronal 

 effects ' never as perfect as with water, however, always showing the tendency to 

 stratification. The characteristic coronas succeeded each other so rapidly that it 

 would be difficult to make them out. Nuclei, however, were still present after 

 over two hours, the eventually white centered coronas showing a continued shrink- 

 age to smaller diameters in accordance with the diminishing number of nuclei 

 present. Twenty exhaustions did not remove them. 



Ilei'e, as above, therefore, the fleeting character of the coronas, their tendency 

 to depait from the normal annular character into stratification, the speed of descent 

 of the color bands,* their I'ise upward on exhaustion like a fog from a lake, are the 

 special characteristics of the colored cloudy condensation occurring in benzine. To 

 these are to be added the strikinef axial coloi-s mentioned alwve. 



5. To explain the above phenomena in their variation from the normal 

 a([ueous corona, it is first necessary to account for the more rapid subsidence of 



' These will be described in a subsequent paragraph. 



' Due both to evaporation at the fog surface as well as to subsidence, very volatile and relatively 

 heavy drops promoting both occurrences. 



