A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 99 



the work, the phosphorus nuclei are at first ionized. Subsidence is without 

 effect. 1 



It is just here that another important question is suggested which may 

 perhaps offer evidence in explanation. It was shown above ( 8, table 2) that 

 the moist air after exhaustion very nearly regains its original temperature after 

 the lapse of even half a minute, while the coronas persist throughout this in- 

 terval and much longer without appreciable change of character. It is difficult 

 to understand why the fog particles do not evaporate proportionately to the 

 rapid rise of temperature, unless there is rapid evaporation and diffusion from 

 the relatively warm inner surface of the walls of the condensation chamber, im- 

 mediately after exhaustion. In any case, the method above used for computing 

 m, the moisture precipitated per cub. cm., will give a result too large, for it 

 takes no account of the rapid increase of temperature in question after the 

 fog particles are produced. The swifter the exhaustion the larger this dis- 

 crepancy (which is probably indeterminable) is liable to be. Thus in the above 

 case for the pressure difference 6p=i'j cm., at 20, the cooling ideally as far as 

 -9.6, rises to 8.8, in consequence of condensation of fog particles, but within 

 |- minute the temperature is nearly 20 again. Hence the precipitated 

 4.6 X io~ 6 g/cm 3 at 8.8 is to remain undisturbed while the moisture content of 

 saturated air rises from about 8.7 X io~ 6 at 8.8 to about 17.2 X io~ 6 (g/cm 3 ) at 

 about 20, leaving an actual deficit of about 4 X io~ 6 g/cm 3 . This would be out 

 of the question unless moisture evaporated immediately from the damp and 

 warmer walls and the pool of water in the bottom of the apparatus to supply 

 the deficiency. But these conditions are vague, for this moisture may actually 

 be precipitated on the colder fog particles. To some extent, therefore, a degree 

 of uncertainty is left in the determination of m, the moisture precipitated per 

 cub. cm., inasmuch as the actual temperature at which the fog particles persist, 

 and to which they have accommodated themselves, is left in doubt. 



It is well to observe, however, that as there will presumably be more evap- 

 oration from the fog particles while the normal air temperature is being 

 regained in case of very rapid than for the slow exhaustion because lower tem- 

 peratures are in general associated with the former case, the discrepancies 

 of plates 12 and 13 above the g-b-p corona may possibly be explained in this 

 way. Compatibly with observation the effect would be less marked as the fog 

 particles are larger. Finally, very large coronas are always more fleeting in 

 character. 



Experimentally and with a bearing on Chapter IX the question is easy of 

 decision. So long as the pressure difference corresponding to the lower limit 

 of spontaneous condensation of moisture from dust -free air is not approached 

 (it will usually be reached at about $p> 20 cm. for the above types of appa- 



' Recent experiments have shown that the very small nuclei associated with larger nuclei evaporate 

 their loads of water after condensation in such a way as to form water nuclei. As the latter must be 

 larger than the original nuclei now held in solution, a reason for the excess of nucleation detected by the 

 instantaneous valve is suggested. 



