74 A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 



(pis the angular radius. The partial exhaustion is to 17 cm. and the standard- 

 ization is by subsidence ( 14). If standardized by diffraction, the w-data 

 would be about .6 smaller, or the upper "green" corona, for instance, showing 

 # = 98,000, would then show n = 60,000 nuclei. The d effect is much smaller, 

 being .2 larger. 



Towards the end of any series the numbers diminish more slowly than the 

 formula requires, but this is naturally the result of the evaporation seen, for 

 instance, in the shrinkage of the coronas. On the other hand, when subsidence 

 is exceptionally rapid, the time during which subsidence takes place, and which 

 can in no case be sharply given (it was not directly timed), is seriously 

 large. On the whole, the agreement is better than was anticipated, and 

 certainly trustworthy. The results coincide in a general way, moreover, with 

 the data found by assuming constant as was done in 5, 9, and the latter 

 equations may be regarded as approximations by expansion of those in this 

 section. 



The graphs, n in terms of s, give evidence in the first series of three cycles, 

 the lower two being merged. In the second series there are apparently four 

 cycles, the two lower being distinct. The horizontal position of the cusps is as 

 closely in accord as the measurements justify. The vertical position suffers 

 from the shift and difficulty surrounding the absolute evaluation of n. Through- 

 out their extent, however, the fundamental similarity of the graphs is unmis- 

 takable, as will be further shown in the corresponding curves for ruby light 

 below. 



Since ' = 6w/W 3 =(6w/7ra 3 ) s 3 = 23 (s/io 4 A) 3 , approximately, the fluctua- 

 tion of n with A is obvious ; but the feature of the phenomenon is none the less 

 the occurrence of cyclic variations in the color of the innermost ring. The correc- 

 tion implied in the last equation would be more than sufficient. The violet 

 coronas are to be depressed as regards n and the red coronas raised in their n 

 values, showing that in the former the measurement referred to red surpassing 

 the last violet, and in the latter to violet beyond the red. It is expedient to 

 state these data in relation to the diameters of the fog particles under observation. 



17. Diameter of fog particles. Having determined the true values of n, 

 the diameters of fog particles may be computed for each aperture, since 



d = -$6m/7m = .02 1~ I/3 . 



The results are given in the tables and are plotted in the corresponding graphs. 

 Each of these (d as a function of 5) shows the three cycles already determined, 

 and the cusps lie at ^ = .0007 to .0008 cm. and ^ = .0005 to .00055 cm., or that 

 the intermediate and particularly luminous cycle covers a range corresponding 

 to about ten times the wave lengths (.00004 to .00008 cm.) of the visible spec- 

 trum. But two of the cusps are unmistakably marked, while in other respects 

 the graphs retain the hyperbolic contour, ds = const. 



Since n~ I/3 is the cubical volume which contains one fog particle, d/n~ I/3 



